Technical Information Flexible Automatic SwitcHed client-Independent Optical Network ASON and its Clients: Optical layer's functionality and transparency Authors: H. Nakajima (Editor), J. Chauvin, K. Mehadji, F. Tillerot (France Telecom R&D) J. Robadey, J.C. Bischoff, S. Burschka, G. Mazza, D. Rodellar, D. Rossier, J. Schneider (Swisscom AG) R. Clemente, G. Ferraris, L. Messina (Telecom Italia Lab SpA) E. Zouganeli, B. Feng (Telenor) V. Andronikidis, T. Doukoglou, S. Drakatou, D. Kapetanakis, N. Konstantinidis, C. Mas Machuca (OTE SA-Intracom) Internal reviewer: T. Doukoglou (OTE SA) Abstract This report complements the previous Technical Information report "Network Requirements and Scenarios" (EDIN: 0199-1012) of project P1012 FASHION and provides a description of Automatically Switched Optical Network (ASON) from the perspective of functionality, management capabilities, grooming, optical transparency and advanced control plane without describing any network solutions. It describes: the functionality from two different perspectives: core-edge and client-server; the specific management requirements for the ASON control plane and a comparison between distributed and centralized management; grooming issues with two options; optical transparency issues from the perspective of functionality, OCh length, protection and wavelength conversion; compatibility of MPS with ASON. EDIN Project 0200-1012 P1012 For full publication April 2003 Disclaimer This document contains material, which is the copyright of certain EURESCOM PARTICIPANTS, and may not be reproduced or copied without permission. All PARTICIPANTS have agreed to full publication of this document. The commercial use of any information contained in this document may require a license from the proprietor of that information. Neither the PARTICIPANTS nor EURESCOM warrant that the information contained in the report is capable of use, or that use of the information is free from risk, and accept no liability for loss or damage suffered by any person using this information. This document has been approved by EURESCOM Board of Governors for distribution to all EURESCOM Shareholders. EURESCOM Technical Information page 3 (93) Table of contents Table of contents................................................................................................................................................ 3 List of Figures .................................................................................................................................................... 7 List of Tables ..................................................................................................................................................... 7 Abbreviations, Acronyms and Terms ................................................................................................................ 8 1 Introduction.............................................................................................................................................. 11 2 Models for ASON .................................................................................................................................... 12 2.1 Short description of OCh Services ................................................................................................... 12 2.2 ASON reference models .................................................................................................................. 12 2.3 Overall description ........................................................................................................................... 13 3 Functionality in the ASON and Client layers .......................................................................................... 15 3.1 Functionality in the core network of ASON .................................................................................... 15 3.2 Functionality at the edge of ASON .................................................................................................. 16 3.2.1 Model for the network boundary.............................................................................................. 16 3.2.2 Functionality related to Client services .................................................................................... 18 3.2.2.1 QoS/CoS compatibility related functionality ....................................................................... 18 3.2.2.2 Resource utilisation oriented ................................................................................................ 19 3.2.2.3 Miscellaneous ...................................................................................................................... 20 3.2.3 Functionality related to optical transport in the core ................................................................ 20 3.2.3.1 Functions for establishing end-to-end optical paths ............................................................. 20 3.2.3.2 Functions for established paths ............................................................................................ 20 3.2.3.3 Functions for Traffic Engineering ........................................................................................ 21 3.2.4 Functionality of the UNI interface ........................................................................................... 21 3.2.5 Functionality related to the physical interfaces ........................................................................ 21 3.2.5.1 Bandwidth grooming functions ............................................................................................ 21 3.2.5.2 Framing adaptation .............................................................................................................. 21 3.3 Server layer functionality ................................................................................................................. 22 3.3.1 Control plane functionality ...................................................................................................... 22 3.3.1.1 Network topology discovery ................................................................................................ 22 3.3.1.2 Optical routing ..................................................................................................................... 22 3.3.1.3 Connection handling functions ............................................................................................ 22 3.3.1.4 Signalling ............................................................................................................................. 22 3.3.1.5 End-to-end protection/restoration ........................................................................................ 22 3.3.1.6 Automatic end-to-end OCh provisioning ............................................................................. 22 3.3.1.7 Node and link management .................................................................................................. 22 3.3.1.8 Policing ................................................................................................................................ 22 3.3.1.9 CAC ..................................................................................................................................... 22 3.3.1.10 QoS handling ................................................................................................................... 23 3.3.1.11 Performance monitoring .................................................................................................. 23 3.3.1.12 Functionality of the UNI interface ................................................................................... 23 3.3.2 OTN plane functionality .......................................................................................................... 24 3.3.2.1 Optical cross-connection ...................................................................................................... 24 3.3.2.2 Optical add-drop .................................................................................................................. 24 3.3.2.3 Traffic grooming .................................................................................................................. 24 3.3.2.4 Wavelength conversion ........................................................................................................ 24 3.3.2.5 Optical multiplex/demultiplexing ........................................................................................ 24 3.3.2.6 Protection ............................................................................................................................. 24 3.3.2.7 Failure detection ................................................................................................................... 24 3.3.2.8 Performance monitoring ...................................................................................................... 25 3.4 Functionality in the client layers ...................................................................................................... 25 3.4.1 Frame Relay ............................................................................................................................. 25 3.4.1.1 Basic functionality of the Frame Relay ................................................................................ 25 3.4.1.2 Functionality of the LMI extensions .................................................................................... 25 3.4.1.3 Functionality related to switched virtual circuits (SVCs) .................................................... 26 3.4.1.4 Functionality related to permanent virtual circuits (PVCs) .................................................. 26 3.4.1.5 Functionality related to overbooking ................................................................................... 26 3.4.2 ATM network........................................................................................................................... 26 3.4.2.1 Generic network functions ................................................................................................... 27 3.4.2.2 Functionality with PNNI protocol ........................................................................................ 27 3.4.2.3 Functionality related to Quality of Service (QoS) ................................................................ 28 3.4.2.4 Routing functionality ........................................................................................................... 28 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 4 (93) EURESCOM Technical Information 3.4.2.5 Functionality related to signalling ........................................................................................ 28 3.4.2.6 Functionality related to survivability ................................................................................... 29 3.4.3 SDH network ........................................................................................................................... 29 3.4.3.1 SDH framing functionality................................................................................................... 29 3.4.3.2 SDH OAM functionality ...................................................................................................... 29 3.4.3.3 Management functionality ................................................................................................... 29 3.4.3.4 Equipment dependent functionality ..................................................................................... 29 3.4.4 IP-based network ..................................................................................................................... 30 3.4.4.1 IPv4and IPv6........................................................................................................................ 30 3.4.4.2 Functionality supported by the conventional best effort IP ................................................. 30 3.4.4.3 Functionality supported by MPLS-TE ................................................................................. 31 3.4.5 Conclusion ............................................................................................................................... 31 3.5 Client-ASON interworking issues ................................................................................................... 32 3.5.1 Basic considerations on Client-ASON interworking ............................................................... 32 3.5.1.1 Functionality in the Client-ASON relationship .................................................................... 32 3.5.1.2 Perturbation schemes for the Client-ASON relationship and interworking ......................... 33 3.5.2 Functionality implementation issues in the control plane ........................................................ 34 3.5.3 Conclusion regarding functionality distribution/modification ................................................. 35 3.6 Conclusion regarding Client layer vs. ASON functionality ............................................................. 35 4 Management requirements ....................................................................................................................... 37 4.1 Introduction...................................................................................................................................... 37 4.2 Survey of management requirements ............................................................................................... 37 4.2.1 Performance management ........................................................................................................ 37 4.2.2 Fault management .................................................................................................................... 38 4.2.3 Accounting management ......................................................................................................... 38 4.2.4 Security management ............................................................................................................... 38 4.2.5 Configuration management ...................................................................................................... 38 4.2.5.1 Network planning and engineering ...................................................................................... 39 4.2.5.2 Installation ........................................................................................................................... 39 4.2.5.3 Service planning and negotiation ......................................................................................... 39 4.2.5.4 Provisioning ......................................................................................................................... 39 4.2.5.5 Status and control ................................................................................................................. 39 4.3 Specific management requirements for the ASON control plane .................................................... 39 4.3.1 Topology management............................................................................................................. 39 4.3.1.1 Node topology management ................................................................................................ 39 4.3.1.2 OCh link topology management .......................................................................................... 40 4.3.1.3 Optical path topology management ..................................................................................... 40 4.3.2 Optical path routing ................................................................................................................. 40 4.3.3 End-to-end path provisioning................................................................................................... 40 4.3.4 Control channel management................................................................................................... 40 4.3.5 Protection management ............................................................................................................ 41 4.3.5.1 Difference between protection and restoration .................................................................... 41 4.3.5.2 End-to-end path protection................................................................................................... 41 4.3.5.3 End-to-end path restoration .................................................................................................. 41 4.3.5.4 Link protection ..................................................................................................................... 41 4.4 Distributed and centralised management ......................................................................................... 41 4.4.1 Distributed management .......................................................................................................... 41 4.4.2 Centralised management .......................................................................................................... 42 4.4.3 Comparison between distributed management and centralised management (Table 3)........... 44 4.4.4 Advantages and disadvantages................................................................................................. 44 4.4.4.1 Centralised management ...................................................................................................... 44 4.4.4.2 Distributed management ...................................................................................................... 45 4.5 Summary .......................................................................................................................................... 45 5 Grooming issues ...................................................................................................................................... 46 5.1 Definition ......................................................................................................................................... 46 5.2 Requirements for grooming function ............................................................................................... 46 5.3 Grooming and services in ASON..................................................................................................... 47 5.3.1 Option: ASON-SDH ................................................................................................................ 47 5.3.2 Option: ASON-OTN ................................................................................................................ 47 5.4 Grooming scenario ........................................................................................................................... 48 5.5 Summary .......................................................................................................................................... 49 6 Optical transparency related issues .......................................................................................................... 50 6.1 Definition of optical transparency.................................................................................................... 50 EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 5 (93) 6.2 Impacts of optical transparency upon the ASON's functionality ..................................................... 50 6.2.1 Functionality implementation in transparent network elements .............................................. 50 6.2.1.1 Transparent optical cross-connect node ............................................................................... 51 6.2.1.2 Optical transparent links ...................................................................................................... 51 6.2.1.3 Functionality implementation issues for transparent network elements............................... 52 6.2.2 Functionality issues in a transparent ASON ............................................................................ 52 6.2.2.1 Degree of influence upon the basic functionality ................................................................. 52 6.2.2.2 Basic functionality issues ..................................................................................................... 53 6.2.2.3 End-to-end functionality issues ............................................................................................ 53 6.2.2.4 Advantages and drawbacks in the functionality ................................................................... 53 6.2.3 Management issues .................................................................................................................. 54 6.2.3.1 Node management................................................................................................................ 54 6.2.3.2 Link topology management.................................................................................................. 54 6.2.3.3 Path topology management .................................................................................................. 54 6.2.3.4 Performance management .................................................................................................... 54 6.2.4 Conclusion ............................................................................................................................... 54 6.3 OCh length, transparent islands, resource allocation ....................................................................... 54 6.3.1 OCh length in a transparent ASON .......................................................................................... 54 6.3.1.1 OCh length issues ................................................................................................................ 54 6.3.1.2 Engineering limitations ........................................................................................................ 55 6.3.1.3 Maximum transmission distances ........................................................................................ 55 6.3.2 Protection/restoration path length issues .................................................................................. 56 6.3.2.1 1+1, M:N protection............................................................................................................. 57 6.3.2.2 Restoration ........................................................................................................................... 57 6.3.3 Transparent islands .................................................................................................................. 58 6.3.3.1 Overcoming OCh length limitation ...................................................................................... 58 6.3.3.2 Optical transparent islands ................................................................................................... 58 6.3.3.3 Hybrid nodes ........................................................................................................................ 58 6.3.4 Resource allocation and wavelength blocking without wavelength conversion ...................... 58 6.3.4.1 Resource allocation .............................................................................................................. 58 6.3.4.2 Wavelength blocking ........................................................................................................... 58 6.4 Signalling network implementation ................................................................................................. 59 6.4.1 Signalling network implementation alternatives ...................................................................... 59 6.4.2 In-band/In-fibre ........................................................................................................................ 59 6.4.2.1 Sub-carrier modulation [11] ................................................................................................. 59 6.4.2.2 Digital wrapper .................................................................................................................... 59 6.4.3 Out-of-band/In-fibre ................................................................................................................ 60 6.4.4 Out-of-band/Out-of-fibre ......................................................................................................... 60 6.4.5 Conclusion ............................................................................................................................... 60 6.5 Wavelength conversion and OXC's properties................................................................................. 61 6.5.1 Network scenario and wavelength conversion alternatives...................................................... 62 6.5.1.1 Wavelength conversion performed by the OLTs ................................................................. 62 6.5.1.2 Wavelength conversion performed by the OXCs................................................................. 63 6.5.1.3 Wavelength conversion performed by dedicated equipment ............................................... 64 6.5.1.4 Conclusion ........................................................................................................................... 66 6.6 Optical transparency in the OTN layer: advantages and drawbacks ................................................ 66 6.6.1 Advantages ............................................................................................................................... 66 6.6.2 Drawbacks ............................................................................................................................... 66 6.7 Conclusion ....................................................................................................................................... 67 7 Compatibility of MPS with ASON ........................................................................................................ 68 7.1 Introduction ...................................................................................................................................... 68 7.2 MPLS ............................................................................................................................................... 68 7.2.1 Label operations ....................................................................................................................... 69 7.2.2 Label allocation ........................................................................................................................ 69 7.2.3 Label distribution and signalling .............................................................................................. 70 7.2.4 Aggregation ............................................................................................................................. 70 7.2.5 Routing..................................................................................................................................... 70 7.2.6 Traffic Engineering .................................................................................................................. 70 7.3 Introduction to GMPLS ................................................................................................................... 71 7.3.1 Main issues involved in the extensions of MPLS to encompass the optical domain (MPS).. 71 7.3.2 Generalised Label .................................................................................................................... 72 7.3.3 Label requests .......................................................................................................................... 73 7.3.4 Generalised LSRs..................................................................................................................... 73 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 6 (93) EURESCOM Technical Information 7.3.5 Architecture models ................................................................................................................. 73 7.3.6 Bi-directional LSPs .................................................................................................................. 73 7.4 On the suitability of MPS for the ASON control plane ................................................................. 74 7.5 Summary .......................................................................................................................................... 74 8 Conclusions ............................................................................................................................................. 75 References ....................................................................................................................................................... 77 Annex A QoS .............................................................................................................................................. 78 A.1 IP Quality of Service ........................................................................................................................ 78 A.2 DiffServ: The Differentiated Services Model .................................................................................. 78 A.3 IntServ: The Integrated Services Model .......................................................................................... 79 Annex B MPLS ........................................................................................................................................... 80 B.1 Multi Protocol Label Switching (MPLS) integration ....................................................................... 80 B.2 MPLS principle: IP packet classification into packet flows............................................................. 80 B.3 MPLS basics .................................................................................................................................... 80 Annex C ATM network .............................................................................................................................. 82 C.1 ATM basics ...................................................................................................................................... 82 C.1.1 ATM switching operation ........................................................................................................ 82 C.1.2 ATM reference model .............................................................................................................. 82 C.2 SVC connection procedure .............................................................................................................. 82 Annex D Generic OTN management requirements (after G.872) ............................................................... 84 D.1 Introduction...................................................................................................................................... 84 D.2 Generic management requirements .................................................................................................. 84 D.2.1 Generic fault, configuration and performance management .................................................... 84 D.2.2 Generic management communications .................................................................................... 84 D.2.3 Generic client/server interaction management ......................................................................... 84 D.2.4 Optical channel layer network management requirements ...................................................... 85 Annex E Optical transparency in the OTN ................................................................................................. 87 E.1 Technology related issues ................................................................................................................ 87 E.1.1 Optical amplification ............................................................................................................... 87 E.1.2 Add-drop multiplexers ............................................................................................................. 87 E.1.3 Wavelength conversion techniques .......................................................................................... 87 E.1.4 Optical cross-connects ............................................................................................................. 87 E.2 Conclusion ....................................................................................................................................... 88 Annex F Requirements on the control plane of ASON .............................................................................. 89 F.1 Introduction...................................................................................................................................... 89 F.2 Routing ............................................................................................................................................ 89 F.2.1 Resource discovery stages ....................................................................................................... 90 F.2.2 Path selection ........................................................................................................................... 91 F.2.3 Transactions ............................................................................................................................. 91 F.3 Logical link/bundle .......................................................................................................................... 92 F.4 Protection ......................................................................................................................................... 92 F.5 Restoration ....................................................................................................................................... 93 F.6 Signalling ......................................................................................................................................... 93 F.7 Control plane failure handling ......................................................................................................... 93 EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 7 (93) List of Figures Figure 1 Logical view of ASON architecture (After ITU Q19/13 [2]). .......................................................... 13 Figure 2 A high level partition of ASON control plane agents (After ITU/COM 15-32-E [4]) ..................... 14 Figure 3 Network boundary scheme ............................................................................................................... 17 Figure 4 Service forwarding scheme at the ingress node. ............................................................................... 18 Figure 5 Distributed management (ASON) .................................................................................................... 42 Figure 6 Centralised management (OTN) ....................................................................................................... 43 Figure 7 Clients between two distant nodes should be groomed several times with other clients in order to use the minimal number of OChs. Without grooming an extra OCh first in red. With a grooming in C and D already opened OCh are used. ....................................................................................................... 46 Figure 8 Grooming the client and client service requirements are taken into account by ASON: this is the ASON-SDH option. ................................................................................................................................. 47 Figure 9 Grooming is managed at the ASON edge. This is the standard ASON or the ASON-OTN. ........... 48 Figure 10 Proposed grooming scenario, by using the SDH. The router and the SDH multiplexer can also have a direct access to a wavelength (in red and blue). G.709 is only shown as one possibility. In this scenario the client interfaces are client-SDH interfaces. .......................................................................... 48 Figure 11 Multi-fibre interconnection scheme. ............................................................................................... 51 Figure 12 WDM interconnection scheme. ...................................................................................................... 51 Figure 13 Network scenario: meshed topology with OXCs and OLTs .......................................................... 62 Figure 14 Wavelength conversion performed by the OLTs ............................................................................ 63 Figure 15 Wavelength conversion performed by the OXCs ........................................................................... 64 Figure 16 Wavelength conversion performed by dedicated equipment .......................................................... 65 Figure 17 MPLS Header ................................................................................................................................. 69 Figure 18 Label swapping example: data that reaches the core LSR with Input Port (IP) 0 and label (IL) 3 will be forwarded to Output Port (OP) 2 with Label (OL) 5. ................................................................... 69 Figure 19 Label hierarchy in GMPLS............................................................................................................. 72 Figure 20 Evolution scenario of vertical network architecture ....................................................................... 74 List of Tables Table 1 Some examples of the ASON functionality and their availability in the client layers. ...................... 33 Table 2 ASON functionality implementation ................................................................................................. 36 Table 3 Comparison between distributed and centralised management ......................................................... 44 Table 4 Comparison of different transmission techniques versus bit-rate and transmission distance, with Erbium doped fibre amplifiers [10]. SMF+DCF stands for a link comprising standard single mode fibre and dispersion compensated fibre. ........................................................................................................... 56 Table 5 signalling implementation and node properties ................................................................................. 61 Table 6 ............................................................................................................................................................. 71 Table 7 Generalised Label request information [14] ...................................................................................... 73 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 8 (93) EURESCOM Technical Information Abbreviations, Acronyms and Terms APS automatic protection switching ASON automatically switched optical network AS-OCh automatically switched OCh ATM asynchronous transfer mode BBER background block error ratio BDR Backup Designated Router BGP Border Gateway Protocol CAC Connection Admission Control CCI Connection Control Interface CIR Committed Information Rate CMIP common management information protocol Colored interface interface with defined wavelength CoS class of service CRC cyclic redundancy check CR-LDP Constraint-based Routing Label Distribution Protocol DCF dispersion compensated fiber DEMUX demultiplexer DiffServ Differentiated Services DR Designated Router DS field DiffServ field DS1 Digital Signal level 1 DSF dispersion shifted fiber DTE data terminal equipment DTL Designated Transit List ESF extended SF ESR Errored Second Ratio FEC Forwarding Equivalence Class GbEth Gigabit Ethernet GMPLS Generalised MPLS G-PID generalised payload identifier Gray interface interface with undefined wavelength IntServ Integrated Services IrDI Inter Domain Interface ITU International Telecommunication Union LDP Label Distribution Protocol LIB label information base EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information LMI local management interface LSP label switched path LSR Label Switched Router MIB management information database MPLS Multiprotocol Label Switching MPS Multi-protocol lambda Switching MS-SPRing multiplex section shared protection ring MUX multiplexer NE network element NEMS network element management system NMI network management interface NMS network management system NNI Node-Node Interface NO network operator OADM optical add-drop multiplexer OAM operation, administration and maintenance OCC Optical Connection Controller OCh optical channel ODSI Optical Domain Service Interconnect OIF Optical Internetworking Forum OLT optical line terminal OMS optical multiplex section OSPF open shortest path first OTN optical transport network OTS optical transport section OVPN Optical Virtual Private Network OXC optical cross-connect PDH plesiochronous digital hierarchy PGL Peer Group Leader PNNI Private Network Node Interface P-OCh permanent OCh PTSE PNNI topology state element PVC permanent virtual circuit (connection) QoS quality of service RAS Reliability, Availability, Survivability RSVP Resource Reservation Protocol SDH synchronous digital hierarchy SESR Severely Errored Second Ratio SF Super Frame 2003 EURESCOM Participants in Project P1012 page 9 (93) EDIN 0200-1012 page 10 (93) EURESCOM Technical Information SLA service-level agreement SMF single mode fiber SNC sub-network connection SNMP simple network management protocol SONET synchronous optical network SPF shortest path first SP-OCh soft permanent OCh SVC switched virtual circuit (connection) TC field type of cost field TCP Transmission Control Protocol TDM time division multiplexing TE Traffic Engineering TMN Telecommunications Management Network TOS type of service TTL Time-To-Live UNI User-Network Interface UPC Usage Parameter Control VC virtual channel VCC virtual channel connection VCI virtual channel indicator VP virtual path VPC virtual path connection VPI virtual path indicator VT Virtual Tributary WDM wavelength division multiplexing EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information 1 page 11 (93) Introduction Automatically switched optical networks (ASON) are based on the so-called overlay model. They feature multiple clients and are foreseen to be deployed mainly in backbone networks. Due to its control plane, which offers automated connection handling capabilities, ASON should provide its clients with a set of new OCh services being described in the previous Technical Information report [1]. The necessary functionality and management capabilities of ASON for providing these new services are presented in this report. This report gives a description of ASON from the perspective of functionality, management capabilities, grooming, optical transparency and advanced control plane, without describing any network solutions. The report starts with a brief overview of OCh services to be provided by ASON (section 1.1). It is followed by a brief description of ASON models currently being discussed in different standardisation bodies (sections 1.2 and 1.3). Functionality in the ASON and its client layers are described in the subsequent chapter 2. ASON functionality is described from two different perspectives: core-edge and client-server. The client functionality is considered in order to introduce a discussion on Client layer-ASON interworking issues. Frame Relay, ATM, SDH and IP are all considered as ASON clients. Chapter 3 is devoted to the description of management requirements. Management requirements related specifically to the ASON control plane are presented together with a comparison between distributed management and conventional centralised management. Grooming is considered in Chapter 4 where two grooming schemes are discussed and a grooming scenario is proposed. Optical transparency is discussed in Chapter 5 with respect to its impacts on the ASON functionality, OCh length issues, signalling network implementation and wavelength conversion. The applicability of MPS to the ASON control plane is discussed in Chapter 6. Finally, a short conclusion is provided. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 12 (93) EURESCOM Technical Information 2 Models for ASON 2.1 Short description of OCh Services In this section, OCh services proposed in [1] are briefly recalled. (a) Permanent OCh service The permanent OCh service (P-OCh service) provides the customer1 with an end-to-end OCh between two end-points. The service is provided by the network operator (NO) on the basis of an agreement between NO and customer. Provisioning of a P-OCh service is the responsibility of the NO, who dedicates multiple network resources to the path. It can be realised via (b) Manual equipment configuration, Centralised TMN-based equipment configuration. Soft-permanent OCh service The soft-permanent OCh service (SP-OCh service) provides the customer with an end-to-end OCh between two end-points. The service is provided by the NO on the basis of an agreement between NO and customer. Provisioning of a SP-OCh service is the responsibility of the NO, who dedicates multiple network resources to the path. It is realised via distributed signalling-based TMNactivated equipment configuration resulting in provisioning speeds faster than for P-OChs. Like for P-OCh, a “natural” understanding of SP-OCh service is that of a “stable in time” service, as an enhanced leased line. (c) Automatically switched OCh service Automatically switched OCh service (AS-OCh service) provides the customer with an end-to-end, real-time activated OChs between two end nodes. The provisioning process is activated by the customer himself, via an user network interface (UNI) signalling interface, when needed only, and completed by means of node-node interface (NNI) signalling. As soon as the AS-OCh service is not necessary anymore it is torn-down. (d) Optical virtual private network service An optical virtual private network service (OVPN service) provides the customer with dedicated network resources (OChs) among customer nodes (two or more). These resources are under the direct control of the customer who can set-up, maintain and tear-down connections on his subnetwork. A common understanding is that an OVPN service is a set of P-OCh services or SP-OCh services for the NO and is generally used by the customer as an AS-OCh service. (e) Lambda trunking service This is an offer of a bundle of P-OChs/SP-OChs between two endpoints; the P-OChs/SP-OChs of the bundle must have the same transmission performances, which implies that they should be routed along the same path and possibly on the same fibre pair. It makes possible a customer to increase the available bandwidth between the endpoint and its flexibility. 2.2 ASON reference models In principle, knowledge of the ASON's structure is not required to describe the functionality. However, it should be useful to look into a few representative models already proposed by different forums in order to have a clear idea about ASON. Some ITU and/or OIF documents can provide basic descriptions for ASON that can serve as a starting point for the discussions. ASON, in providing multi-client services, is characterised by a control plane that manages network elements in the optical transport network plane (Figure 1). This description of ITU Q19/13 [2] is In this document – unless otherwise stated –"customer" refers to both an end customer and a client layer for ASON. 1 EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 13 (93) not far from the description provided by OIF [3]. In the next section, a brief description of ASON is being provided. Network Management System ASON control plane User signaling OCC OCC OCC NNI NMI OCC IrDI_NNI UNIcontrol CCI Clients e.g. IP, ATM, TDM Switch Switch Switch IrDI UNIData Optical Transport Network Figure 1 Logical view of ASON architecture (After ITU Q19/13 [2]). OCC: Optical connection controller; UNI: User network interface; CCI: Connection control interface; NNI: ASON control node-node interface; IrDI: Inter domain interface; NMI: Network management interface. 2.3 Overall description The ASON is intended to allow switching of (optical) network connections within the Optical Transport Network (OTN) under control of its own signalling network (Figure 1). ASON definition implies the existence of three separate planes in the network: the optical transport plane which provides the functionality required for the transport of the client signals of the ASON; in particular, it provides the capability to cross-connect the characteristic information of the optical channels; the ASON control plane which provides the functionality required for establishing end-to-end connections of client signals with the properties (in terms of protections applied, duration and time scheduling of the connection, etc.) that are specified by the customer himself during connection set-up phase; the management plane which performs management functionality both related to the transport plane and to the control plane. In addition to these planes, ASON could have interfaces as follows: UNI: User network interface; CCI: Connection control interface; NNI: ASON control node-node interface; IrDI: Inter domain interface; NMI: Network management interface. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 14 (93) EURESCOM Technical Information This representation, rather succinct, can be detailed by introducing ASON control plane agents performing specific functions. Figure 2 shows an example of partition of agents in different parts of ASON. User control plane agents S-UNI Service provider Control Plane SP CP agents Sig CAC Rtg SP CP agents Link mgmt Dir svc & Name translation Sig CAC S-NNI Link mgmt S-UNI Rtg User NE IrDI/Pre-OTN O-UNI Service provider Control Plane S-NNI Link Mgmt Surv Traffic Policing Link Disc Service Provider NE Link Mgmt Surv Traffic Policing Link Disc Link Mgmt Surv IrDI/ IaDI O-NNI Link Mgmt S-NNI Surv Traffic Policing Link Disc Traffic Policing Link Disc S-UNI/NNI - Signaling UNI/NNI O-UNI/NNI - Optical UNI/NNI Figure 2 A high level partition of ASON control plane agents (After ITU/COM 15-32-E [4]) User NE: User Network Element; Sig.: Signalling; Link Mgmt: Link management; Rtg: Routing; Dir svc & Name translation: Directory service and Name translation; Surv: Survivability; Link Disc: Link discovery; IaDI: Intra-domain interface. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information 3 page 15 (93) Functionality in the ASON and Client layers ASON is based on the overlay model and hence should provide its different Clients with a set of OCh services defined in [1]. The functionality is described from two different perspectives: coreedge and client-server. In general, it is difficult to distinguish physically the core and edge of a network since some client devices are often connected to core nodes that act as edge node at the same time. This means that a network element can include both edge and core functionality. Then, the core-edge perspective refers to a picture of the network for a given connection. In this case, one can always define the edges and the core of network as follows: the end points and transit nodes of the connection correspond to the edge and core nodes respectively. 3.1 Functionality in the core network of ASON The core functionality of an ASON is mainly associated with switching of optical channels (carried by wavelengths) within the network. This is, regardless if this action is triggered by the management plane or directly by the Client Network using signalling through the UNI. Core functionality is typically supported by signalling between the network nodes through the NNI. The core ASON functionality is listed below. Network topology discovery (or resource discovery) This functionality provides routing protocols with enough information about the network topology (and resources) for doing route computation. It includes the ability to learn how Network Elements are physically connected together. The information is exchanged through NNI and stored in a database at each network element. Optical routing This functionality allows finding an optical path from the source to the destination node with some constraints. It requires the knowledge of the network topology and a set of constraints to be applied to. Signalling A set of functions associated with the signalling network (or control channels) that allows the exchange of signalling and management messages through specific interfaces between NEs, between the NMS and NEs or between ASON and its Clients. Signalling includes connection-handling functionality as well: Connection set-up This functionality allows establishment of a new optical path connecting two nodes in the network. This is done by a signalling protocol in which messages are exchanged through NNI between the involved nodes. Connection tear-down This functionality allows deletion of an existing optical path. It can be done by a signalling protocol. Connection modification This functionality allows modification of some characteristics of an existing optical path connecting two nodes in the network. Some of the parameters associated to a connection that could be changed are: frame format, bandwidth, type of protection/restoration and priority (if applicable), etc. End-to-end protection/restoration This is a set of functions involved in the recovery of an end-to-end optical connection affected by a failure. It includes failure detection, path teardown, optical routing, path set- 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 16 (93) EURESCOM Technical Information up functions to be used to perform, in particular, automatic end-to-end protection and automatic end-to-end restoration. End-to-end path protection often uses a diversely routed path (1+1 type protection), typically allocated during the connection set-up, and when the failure occurs traffic is simply switched from the working path to the protection path. In case of restoration (standard reroute), the failure triggers a rerouting process that is composed of teardown of the failed path, computation of a new path and set-up of the new path. This process involves routing and signalling functions. Automatic end-to-end OCh provisioning This functionality allows automatic provision of an end-to-end OCh and is composed of: An OCh request by NMS or Client via UNI; Route computation (done by optical routing); OCh set-up including resource reservation, cross-connect configuration (done by signalling). Node and link management This refers to a set of functions required for managing a node and its links. Node management consists in managing the node attributes such as port states, switching capabilities. Link management consists of managing the link attributes such as the link state, the link capacity. The information required for their management is stored in a local management information base (MIB). Policing This is a set of functions for providing with policy-based management to deploy policies in various domains such as QoS, security, service management and network resource utilisation. In particular, policing functions are necessary to regulate and control the information propagation across interfaces (UNI, NNI) and the actions allowed on behalf of the users. Policy is implemented in each node (at the edge as well as in the core). CAC Connection admission control could be implemented in the core. 3.2 Functionality at the edge of ASON 3.2.1 Model for the network boundary Network edge nodes (EN) are not only incoming and outgoing points for client traffic streams but also endpoints of optical paths (or channels). Client traffic streams issued from edge devices, like IP routers, ATM switches, SDH cross-connects, Frame Relay (FR) equipment, are forwarded to the network edge nodes (ingress EN). On the other hand, inter-nodes (core) traffic is carried by optical channels (or an optical path linking the ingress edge node to the egress one) and terminates at the edge nodes. The NMS (network management system) gives, according to the service-level agreement (SLA), instructions telling how client traffic streams have to be handled (forwarding, extraction, drop) at the edge nodes (Figure 3). EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 17 (93) NMS SLA IP EN ATM EN OCh Services EN Ingress SDH EN FR EN Figure 3 Network boundary scheme The edge devices deliver various client traffic streams with different QoS requirements2 whereas a set of optical path services are provided in the core of the network. Clearly, the traffic streams on the outside of the network have to be matched to those provided by optical path services to be transported over the network. The traffic streams having exactly the same characteristics as one of transport (core) services could directly be forwarded whereas those which do not no match with any of the transport services should be rejected. When their traffic characteristics are partially matched they should have to be classified and conditioned in order to be grouped (Figure 4). The functionality we are here dealing with is mainly how to match a client traffic stream with a core service stream. Indeed, those operations should have to be made according to the SLA. The scopes to be considered at the network boundary nodes should be: Functionality related to the client services The user edge devices connected to the edge nodes deliver a variety of service requests in terms of QoS. If there are some transport services partially matched to the client service request, the request should have to be adapted to the available transport services according to SLA in terms of QoS. In some cases, the client could request a custom service if the network resources are available. Functionality related to the core network Some functions relating to end-to-end optical paths should have to be performed at the edge node. For instance, the ingress edge node should initiate most of the functions for establishing (or withdrawing) an end-to-end path (or OCh) between the ingress and egress node. It also should provide core nodes with information necessary to establish (or withdraw) a path. The information being specific to the client includes e.g. the number of channels. Functionality related to the physical interfaces The adaptation should address the framing and bandwidth between the user edge devices and the ASON's boundary nodes. The input/output ports of a user edge device have to be connected to the add/drop ports of edge nodes with the right framing and bandwidth. When the client signal bandwidth does not match to the bandwidth of any available port, grooming has to be performed unless the client bandwidth exceeds the maximum bandwidth available at add/drop ports. 2 QoS requirements are not defined the same way in the different layers. SDH requirements are related to performance and protection while IP and ATM requirements include delay, jitter, bit rate, etc. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 18 (93) EURESCOM Technical Information Ingress node Forwarding category Clients Exactly matched Same CoS end-to-end Same CoS per link QoS/CoS undefined Custom service DROP Figure 4 Service forwarding scheme at the ingress node. 3.2.2 Functionality related to Client services The user edge devices connected to the edge nodes can deliver a variety of service requests in terms of QoS3. These requests have to be compatible with SLA and the services available in the core network. The adaptation can involve the QoS/CoS compatibility as well as network resource utilisation. Whether the connection request stems from the network operator or a client could also ask different functions. QoS parameters could be: Service availability: the reliability of the connection including priority, survivability, time to connection, recovery time, etc. Protection method Priority Throughput Bit error rate Delay & Delay variation A set of features and characteristics defining CoS: Group values of QoS parameters can define pre-agreed Class of Service (CoS). 3.2.2.1 QoS/CoS compatibility related functionality Each traffic stream delivered by a user edge device has specific QoS requirements that are set by the application being carried. There are functions related to classifying, checking the compatibility up and probably conditioning the traffic streams. These functions prepare client traffic streams for forwarding. Functions for QoS/CoS parameter checking; 3 It should be necessary to define QoS/CoS parameters for the transport services and client traffic streams. However this problem is here out of the scope. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 19 (93) A set of functions needed for checking QoS/CoS parameters of client traffic streams. Functions for QoS/CoS classification; A set of functions needed for classification of client traffic streams as a function of QoS/CoS. There will be a QoS/CoS undefined class. Functions for QoS/CoS compatibility search; A set of functions for sorting out the least common based (or another criteria) QoS/CoS parameters among the client traffic streams. Functions for QoS/CoS conditioning; A set of functions needed for conditioning. They are metering, marking, shaping, and policing of client traffic streams. These are pertinent within IP-DiffServ. 3.2.2.2 Resource utilisation oriented In order to optimise the network resource utilisation, traffic streams requiring the same QoS have to be grouped and routed over a common path. This procedure includes the case in which a common path could be an OCh shared with a set of traffic streams defined on the least common based QoS parameters. Each class of traffic streams having the same characteristics is mapped into a class of transport services. Service bundling function; Some traffic streams of the same service type can be transported over a same path or a same link (OCh, between two adjacent nodes). Such services are bundled at the ingress node. This function bundles services as are suggested from the compatibility search. It can be used to bundle several OChs or a working and a protection path. CoS (and/or QoS) grouping function; Some traffic streams of the same CoS/QoS can be transported over a same path or a same link. Such services are grouped at the ingress node. CoS/QoS grouped streams form Forwarding Equivalence Class (FEC) as in IP domain. Functions for obtaining current status information of network resources; A set of functions needed for obtaining information on currently available network resources. Core service compatibility function; Each bundled services or CoS/QoS grouped streams is asked for finding core services compatible with them. This establishes a table of compatibility with the transport (core) services. Mapping functions; A set of functions needed to map bundled services or CoS/QoS grouped streams to the available transport services. This mapping could be end-to-end or per link. Undefined CoS/QoS streams could be mapped into available transport services. Service and end-device discovery functions; A set of functions needed for discovering end-to-end services provided between (or among) interfaces of user end devices. Transport service request functions; A set of functions enabling a client to request for network resources. The validity of the request has to be verified by CAC before the activation of connection. Automatically switched OCh (bandwidth-on-demand) and automatically switched OVPN services are possible services to be offered. Custom service request function; 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 20 (93) EURESCOM Technical Information If the network resources are available, according to SLA, a client could request a custom-made transport service. 3.2.2.3 Miscellaneous Connection Admission Control function; It checks the coherence between SLA and the request as well as the network resources availability. This is mandatory for client (bandwidth) request services. Connection priority function; When receiving simultaneously more than two connection requests to the same destination port (or at the same input port), ASON has to select a request as a function of priority assigned to the traffic stream by SLA. This function sorts a priority traffic out among several traffic streams. This could be applied to both ingress and egress nodes. QoS monitoring function; This function measures QoS (via parameters), compares with that defined in the SLA and judges the compliance. Accounting functions; A set of functions that acquire and store all the necessary information (OCh up Time, OCh Capacity, Traffic QoS/CoS type etc) associated with Client Requests (via the UNI). The values of these parameters are then fed to each NO billing system for processing. 3.2.3 Functionality related to optical transport in the core Since the edge nodes are endpoints of optical paths (or channels) of the ASON, there are functions specifically related to the setting-up and withdrawing of optical transport services that are initiated by the edge nodes. For instance, when the NMS (or client via the UNI interface in some cases) requests for provisioning a new end-to-end optical path (or OCh) over the core network, the ingress node should initiate such procedure. 3.2.3.1 Functions for establishing end-to-end optical paths Functions for provisioning an optical path; A set of functions are needed for initiating at the ingress node automatic provisioning of an end-to-end optical path to the egress node. Functions for providing core nodes with information necessary to establish a path. Path set-up function; This function initiates path set-up procedure at the ingress node according to the request of the NMS (or client). Functions for resource discovery; A set of functions needed for discovering available resources of the network. Functions for network topology discovery; A set of functions needed for discovering the network topology. Path calculation function. This function, implemented in the ingress nodes, determines the best path according to its database. 3.2.3.2 Functions for established paths Functions for end-to-end protection activated at the ingress node; EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 21 (93) A set of functions needed for activating end-to-end path protection at the ingress node. Functions for restoration by reroute; A set of functions needed for activating end-to-end path restoration by reroute at the ingress node. Functions for obtaining current status information of the path; A set of functions needed for retrieval of current status information of the path to the ingress node. Functions for path parameters changing; A set of functions needed for changing path parameters at the ingress node. Functions for the path teardown. A set of functions needed for performing the teardown of an end-to-end path from the ingress node. 3.2.3.3 Functions for Traffic Engineering Functions for optimisation of resource utilisation; A set of functions needed for the optimisation of resource utilisation initiated at the ingress node. One function is load balancing of bundled end-to-end paths. 3.2.4 Functionality of the UNI interface ASON should implement signalling functionality (or protocol) common with client layers (see section 2.3.1.12). 3.2.5 Functionality related to the physical interfaces The different clients considered (i.e. IP, SDH, ATM, FR) can use a variety of interfaces to connect to edge nodes of ASON. The framing and bandwidth of these client interfaces can be different from those of the edge nodes. Therefore the bandwidth grooming and frame adaptation functions are required at the edge nodes. This is independent of the optimisation of resource utilisation. 3.2.5.1 Bandwidth grooming functions Typical bandwidth of OCh is 2.5 Gb/s (or 10Gb/s). Some commercial OXCs have built-in grooming functions with the bandwidths ranging from 155Mbps to 2.5 Gbps, determined by interface cards. If traffic is groomed at the edge nodes, grooming functions are also required at the core nodes (point-to-point). Function to split traffic into two (or more) groomed paths at the ingress and to combine them at the egress node. SDH mux/demux functions. Function to groom several traffic streams of same FEC. For fully transparent nodes, there is no possibility for grooming in the nodes unless it is performed before the feeding to the nodes. No need for grooming for equipment having full bit-rate interfaces to the edge node. 3.2.5.2 Framing adaptation Typical OXCs or OADMs use SONET/SDH interface ports. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 22 (93) EURESCOM Technical Information The user edge devices have to use built-in SONET/SDH interfaces or frame adaptor to SONET/SDH. Most of IP routers, SDH cross-connects, ATM switches as well as GbE switches have SONET/SDH framed interfaces. Frame adapter function: for a transparent node4 no frame adaptation is required. 3.3 Server layer functionality ASON is seen as the server of the client layers. According to the two-layer view of ASON, some functions could be performed in the control plane and some in the OTN plane or in both. This section describes how the functionality could be distributed between the two planes. 3.3.1 Control plane functionality An ASON differs from a traditional OTN by the existence of a control plane. The ASON control plane provides the functionality required for handling end-to-end optical channels and connection requests and managing network elements. 3.3.1.1 Network topology discovery See section 2.1. 3.3.1.2 Optical routing See section 2.1. 3.3.1.3 Connection handling functions See section 2.1. 3.3.1.4 Signalling See section 2.1. 3.3.1.5 End-to-end protection/restoration See section 2.1. 3.3.1.6 Automatic end-to-end OCh provisioning See section 2.1. 3.3.1.7 Node and link management See section 2.1. 3.3.1.8 Policing See section 2.1. 3.3.1.9 CAC See section 2.2.2.3. 4 At this point, the transparent node is not well defined. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information 3.3.1.10 page 23 (93) QoS handling See section 2.2.2. 3.3.1.11 Performance monitoring Its basic function is to track system, network or service activities in order to gather the appropriate data for determining performance. This functionality is an integral part of the management, policing, CAC, and/or QoS handling activities. Only a part of this functionality involves the OTN layer. 3.3.1.12 Functionality of the UNI interface They are typically supported by the signalling at the UNI interface. Service discovery; It allows the Client NE connected to ASON through the UNI to discover which transport services is available over ASON with a particular Client NE. Address assignment; The ASON could automatically provide an address to the Client NEs connected to the different UNI interfaces. Security; Security capabilities should be available in an ASON in order to control which action can be performed by a certain Client Network Element connected to the UNI. For example, a customer equipment could have different access to the Optical Network resources that an equipment owned by the same Network Operator of the ASON. Security capability should include the authentication of the Client Network Element when it is connected to the UNI and the association to a certain Client Profile that describes which requests it is allowed to perform. Connection request handling through the UNI; The requests coming from the Client Layer through the UNI should be processed by the Optical Network Element and the corresponding actions should be triggered if the Client Profile associated to the Client Network Element originating those requests allows them. Some basic requests are listed below. Connection set-up request; It is the request through the UNI that triggers the set-up a new connection. It is based on signalling on the UNI and the request can include a number of information about the characteristics of the required connection such as: frame format, bandwidth, type of protection/restoration and priority (if applicable), etc. The connection request triggers the Connection Set-up process and, at the end of it, the ASON Network Element should notify to the Client the result of the Connection set-up (success or failure). Connection teardown request; It is the request through the UNI that triggers the teardown of an existing optical path connecting two nodes in the network. Connection modification request; It is the requests through the UNI that triggers the modification of some of the parameters of an existing connection without the need of deleting it and setting it up again. It should include the parameters of the connection that should be modified. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 24 (93) 3.3.2 EURESCOM Technical Information OTN plane functionality Functionality provided at the OTN plane level is mainly related to the cross-connection of optical channels and failure detection. These are native functions of the OTN layer. 3.3.2.1 Optical cross-connection Optical cross-connection cross-connects optical channels and could be: Port-to-port (or fibre-to-fibre); Wavelength-to-wavelength; Or both. 3.3.2.2 Optical add-drop This functionality allows inserting and/or extracting optical channels or signals at an OCh rate or sub-OCh rate. 3.3.2.3 Traffic grooming This functionality allows handling traffic flows at sub-data-rates of an OCh and in general requires a digital cross-connect core. It also allows filling an OCh up with a number of sub-OCh rate flows and should be performed at the edge of ASON. 3.3.2.4 Wavelength conversion This functionality allows converting a wavelength to one another by using a fully optical or O/E/O conversion technique and is optional for transparent nodes to save the consumption of fibres or wavelengths. 3.3.2.5 Optical multiplex/demultiplexing This OMS level functionality is required to transport several wavelengths in one fibre. 3.3.2.6 Protection At OTN level, protection can be at the network element (NE) level: Link level; Line card level; Fabric level. In the case of SDH interface cards, Automatic Protection Switching (APS) is also available. 3.3.2.7 Failure detection This functionality performs the detection of failures at: the link level (fibre cut, disconnection, …); the equipment level (line cards, …); the fabric level. and propagates alarm to the management system. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information 3.3.2.8 page 25 (93) Performance monitoring Performance monitoring that aims at evaluating the performance of an OCh should be carried out at the ingress or egress nodes. This information is necessary to verify the conformance with SLA. 3.4 Functionality in the client layers Functionality in the client layers is briefly described in this section. The considered clients are Frame Relay, ATM, SDH and IP. 3.4.1 Frame Relay Frame Relay is a packet switched, connection oriented technology that operates at the physical and data link layers. It provides a means for statistically multiplexing many logical data conversations (referred to as virtual circuits) over a single physical transmission link. The Frame Relay packets, called frames, are of variable length and are switched across the network devices regardless of their content in an effectively transparent manner resembling a leased line service. The Frame Relay specifications define a user data exchange plane (U-plane) where only the Physical Layer and a subset of the Data Link Layer is used and a control plane (C-plane, in-channel signalling layer-3 messages) for signalling. Frame Relay provides users with permanent virtual circuits (PVCs) and switched virtual circuits (SVCs). 3.4.1.1 Basic functionality of the Frame Relay Frame delimitation, alignment and transparency; Frame multiplexing and de-multiplexing; Bit error detection (CRC); Congestion notification functions; Signalling functions for handling PVCs and SVCs. 3.4.1.2 Functionality of the LMI extensions The local management interface (LMI) extensions make supporting large, complex internetworks easier. LMI functions are as follows: Virtual circuit status messages; Provide communication and synchronisation between the network and the user device, periodically reporting the existence of new PVCs and the deletion of already existing PVCs and generally providing information about PVC integrity. Multicasting; Allows a sender to transmit a single frame but have it delivered by the network to multiple recipients. Multicasting supports efficiently the sending of routing protocol messages and address resolution procedures. Global addressing; Gives connection identifiers global rather than local significance, allowing them to be used to identify a specific interface to the Frame Relay network. Simple flow control; Provides for an XON/XOFF flow control mechanism that applies to the entire Frame Relay interface. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 26 (93) 3.4.1.3 EURESCOM Technical Information Functionality related to switched virtual circuits (SVCs) Switched virtual circuits (SVCs) are temporary connections used in situations requiring only sporadic data transfer between data terminal equipment (DTE) devices across the Frame Relay network. Functions related to a communication session across an SVC are as follows: Call Set-up function: A function for establishing a virtual circuit between two Frame Relay DTE devices. Data Transfer function: A function for ensuring the transmission of data between the DTE devices over the virtual circuit. Connection maintaining function (Idle): A function for maintaining the connection between DTE devices active while no data is transferred. If an SVC remains in an idle state for a defined period of time, the call can be terminated. Call Termination function: A function for terminating the virtual circuit between DTE devices. 3.4.1.4 Functionality related to permanent virtual circuits (PVCs) Permanent virtual circuits (PVCs) are permanently established connections that are used for frequent and consistent data transfers between DTE devices across the Frame Relay network. Communication across a PVC does not require the call set-up and termination. Specific functions are: Data Transfer function: A function for ensuring the transmission of data between the DTE devices over the virtual circuit. Connection maintaining function (Idle): A function for maintaining the connection between DTE devices active while no data is transferred. 3.4.1.5 Functionality related to overbooking Functions for overbooking of the service: Bursty frame function. Excess frames offered to the network in excess of CIR being admitted to the network if there is available capacity. These excess frames will be marked (DE bit, Discard Eligible), to be the first to drop in case of congestion. The Committed Information Rate (CIR) is the transport speed, which the Frame Relay network will maintain between service locations. 3.4.2 ATM network ATM provides for a connection-oriented cell transport service. The ATM packets called cells are of fixed length (53 bytes). The virtual connections of the ATM fall into two different levels. Virtual Paths (VPs) and Virtual Channels (VCs). VC is the most basic type while a VP can contain multiple VCs. Each VP/VC is characterised by an identifier (VPI/VCI) on which the switching of the cells is based. ATM connections are unidirectional or bi-directional point-to-point or unidirectional point-to-multipoint. ATM provides three types of services: permanent virtual connection (PVC), switched virtual connection (SVC) and connectionless service. A PVC allows direct connectivity between sites. Similar to a leased line, a PVC guarantees availability of a connection and does not require call setup procedures between switches. A SVC is created and released dynamically and remains in use only as long as data are being transferred. Dynamic call control requires a signalling protocol between the ATM endpoint and the ATM switch. The PVC's are established by the network administrator while the SVC's upon request from an ATM end device. ATM offers Quality of Service (QoS) introducing service categories. The ATM Service Category relates quality requirements and traffic characteristics to network behaviour (procedures and EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 27 (93) parameters). It is intended to specify a combination of QoS commitment and traffic parameters that is suitable for a given set of applications (user interpretation) and that allows for specific multiplexing schemes at the ATM layer (network interpretation). A Service Category used on a given ATM connection, among those that are made available by the network, has to be implicitly or explicitly declared at connection set-up. All service categories apply to both Virtual Channel Connections (VCCs) and Virtual Path Connections (VPCs). 3.4.2.1 Generic network functions Connection Admission Control (CAC); It is defined as the set of actions taken by the network during the call (virtual connection) setup phase, or during call re-negotiation phase, to determine whether a connection request can be accepted or rejected. Network resources (port bandwidth and buffer space) are reserved to the incoming connection at each switching element traversed, if so required, by the service category. A generic CAC is needed in the path selection process to determine if a link or node is likely to have enough resources available to support the proposed connection. Usage Parameter Control (UPC) or Policing; It is defined as the set of actions taken by the network to monitor and control the traffic offered and the validity of the ATM connection at the User to Network Interface (UNI). The main purpose of UPC is to protect network resources from malicious and unintentional misbehaviour, which can affect QoS of other already established connections. Violations of negotiated parameters are detected and appropriate actions can be taken (e.g. cell tagging, discard). Feedback Controls; They are defined as the set of actions taken by the network and by the end-systems (possibly cooperating) to regulate the traffic submitted on ATM connections according to the state of network elements. Specific Feedback Control procedures may be associated with a service category. 3.4.2.2 Functionality with PNNI protocol The following functionality are identified in the case of PNNI protocol: QoS; A set of functions for providing QoS guarantees. Routing (source routing); A set of functions for discovering the network topology and for synchronising the nodal information, topology state information and reachability information. A set of functions for selecting routes to the destination node. Signalling; A set of functions for handling a connection request (establishing, releasing, reestablishing). Switching; A set of functions for switching ATM cells from an input port to an appropriate output port according to VPI/VCI. Survivability; A set of functions for protecting virtual connections. Management. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 28 (93) 3.4.2.3 EURESCOM Technical Information Functionality related to Quality of Service (QoS) ATM supports quality of service (QoS) guarantees comprised of traffic contract, traffic shaping and traffic policing. A traffic contract specifies the intended data flow in terms of values for peak bandwidth, average sustained bandwidth and burst size among others. When an ATM end-system connects to an ATM network, it enters a contract with the network based on QoS parameters. Traffic shaping is the use of queues to constrain data bursts, limit peak data rate, and smooth jitters so that traffic will fit within the promised envelope. ATM devices are responsible for adhering to the contract by means of traffic shaping. Traffic policing can be used for enforcing the contract. ATM switches can measure the actual traffic flow and compare it with the agreed-upon traffic envelope. If the switch finds that traffic is outside of the envelope, it can set the CLP bit for allowing any switch to drop it during periods of congestion. 3.4.2.4 Routing functionality The functions of the PNNI routing protocol include: Discovery of neighbours and link status (via routing control channel); Synchronisation of the nodal information, topology state information (PTSE), reachability information; Flooding of PTSEs; Election of peer group leaders (PGLs); Summarisation of topology state information; Construction of the routing hierarchy; Path selection; Constraint-based routing at the VC level; Support for administratively configurable explicit VC paths; 3.4.2.5 Functionality related to signalling ATM signalling is a set of protocols used for call/connection establishment and clearing over ATM interfaces (UNI, NNI). It can handle (via Associated signalling) a connection request along the path computed by routing protocol. It supports: Associated signalling; By default, the signalling channel on VPI=0 controls all of the virtual paths on the physical interface. Designated Transit Lists (DTL); They are used to carry hierarchically-complete source routes. A DTL is a complete path across a peer group, consisting of a sequence of Node IDs and optionally Port IDs traversing the peer group. Crank back; When the call cannot be processed according to the DTL, it is cranked back to the creator of that DTL with an indication of the problem. This node may choose an alternative path over which to progress the call or may further crank back the call. Soft permanent virtual path connections (PVPCs) and permanent virtual channel connections (PVCCs). EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 29 (93) A soft PVPC or PVCC is one where the establishment within the network is done by signalling. By configuration the switching system at one end of the soft PVPC or PVCC initiates the signalling for this. 3.4.2.6 Functionality related to survivability A function related to the survivability of VCs is reroute that sets-up an alternate routing of call setup in case of connection set-up failure. 3.4.3 SDH network Synchronous Digital Hierarchy (SDH) standard, as the Synchronous Optical Network (SONET), is based on the principles of direct synchronous multiplexing in which individual tributary signals can be multiplexed directly, by using synchronous transport frames, into a higher-rate synchronous signal without intermediate stages. The signal structure provides built-in signal capacity (overhead) for advanced network management and maintenance at the Path, Multiplexer and Regeneration levels. These capabilities are mainly based on the framing and OAM functionality. 3.4.3.1 SDH framing functionality The following functionality is concerned with payload adaptation to the frame size. Byte-interleaved multiplexing; Tributary unit framing; Concatenation; Payload pointer action; Frame alignment; Clock synchronisation; 3.4.3.2 SDH OAM functionality The SDH network is capable of generating alarm and performance monitoring data and propagating this information throughout the network. The following functional components are involved: Continuity testing of Path and between sections; Error monitoring; Error monitoring checks the quality of signal. It is performed on the Path, Multiplexer and Regeneration sections. Status and performance monitoring; Alarm generation, detection and propagation; Automatic Protection Switching (APS) at the Path and Multiplexer section levels; 3.4.3.3 Management functionality NMS manages connection set-up and teardown through the network element management system (NEMS) managing the network elements attached. 3.4.3.4 Equipment dependent functionality Add and drop functions; These are carried out using some of the framing functionality. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 30 (93) EURESCOM Technical Information Digital cross-connection; 3.4.4 IP-based network 3.4.4.1 IPv4and IPv6 An IP-based network uses IP routers to forward datagrams containing information from the source to the destination without a connection (connectionless feature) over the network. The client devices are IP routers that perform a set of functions required for datagram handling. At present there are two versions of IP protocol: IPv4 and IPv6. The functionality provided by these versions is significantly different. In particular, thanks to a 16 bytes long addressing field, IPv6 can offer a new multicast address format and introduces the new anycast address. It supports QoS at the network layer level. Especially, when used together with protocols for network resource reservation like RSVP, real end-to-end QoS may be provided using the IPv6 functionality. By introducing the fixed header format, IPv6 provides hardware processing of the header leading to speed up of the processing. In addition, the fragmentation of the data is done at the source and not as in IPv4 by the router having the limiting packet size capability. Associated with this change the IPv6 routers must, as a minimum, support datagrams of 576 bytes, instead of 68 bytes required for IPv4 routers. All information on the fragmentation is moved from the IP header to an extension header in order to simplify the protocol and speed up the processing of IP datagrams in the routers. In the following, we look at the IPv4 functionality. 3.4.4.2 Functionality supported by the conventional best effort IP Conventional IP running on a link-state protocol supports the following functionality: Topology discovery; This is a set of functions leading to establish a link-state database. It includes adjacency discovery, DR and BDR election, database synchronisation. Route selection; Functionality required for generating a routing table on the basis of the link-state database. Routing information maintaining; A set of functions needed for keeping update the routing table. Hop-by-hop forwarding; Packet fragmentation and reassembly; Reroute restoration; This is a set of functions for redirecting the traffic (datagrams) into an alternative route (when this exists) and involves adjacency discovery and updating the routing table. IP restoration can recover from any type of faults (cable cuts, transmission equipment/line card failures, router failures etc.). IP protection; IP level protection is a error check-sum of the header information within each datagram. Quality of Service. Classical IPv4 cannot support QoS unless using the Differentiated Services (DiffServ) or Integrated Services (IntServ/RSVP) architectures [Annex A]. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information 3.4.4.3 page 31 (93) Functionality supported by MPLS-TE Multiprotocol label switching (MPLS) is by definition protocol independent. An IP-based network is then one of its Clients. Connection-like aspects of MPLS allow traffic engineering (TE) capabilities for IP. MPLS protocol with TE extensions [5] provides the functionality such as: Label switched path (LSP) handling functions; Functionality required for setting-up, tear downing, maintaining an LSP. Label distribution functions; A set of functions for performing label binding, resource reservation, … Forwarding function; Label handling functions; It includes Label swapping, stacking, popping, … Constraint-based routing functions; A set of functions needed for finding the best routes on the basis of a set of criteria. TE functionality; A set of functions needed for performing CoS routing, QoS differentiation, resource management, modification of path attributes etc. Trunk admission control; It determines if resources are available, may teardown trunks with a lower priority, does local accounting. End-to-end LSP protection/restoration; A set of functions for performing end-to-end protection and restoration of an LSP. They include the functionality such as LSP setting-up, tear downing, routing, … Path maintenance functionality. It monitors status of established route, initiates rerouting in the presence of failures, looks for opportunities to re-optimise route. Performance monitoring; Capability to obtain statistics at the traffic trunk level. Traffic shaping; Policing. 3.4.5 Conclusion One key feature of ASON is providing different Clients with OCh services defined in [1]. This postulate means ASON should simultaneously cope with different signal types, connection types, signalling structures, QoS requirements, etc. From the point of view of the functionality, ASON could deal with a variety of functionality implemented at the client layers. A short study on Frame Relay, ATM, SDH and IP based networks reveals: Most of functionality is Client specific. Some functionality is common for several Clients. For instance, PNNI and MPLS-TE have several common functionality like constraint-based routing, signalling. ATM working within PNNI protocol has a complete set of functions for providing with PVC/SVC services very similar to the ASON OCh services. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 32 (93) EURESCOM Technical Information Functionality provided by the IP based network are strongly dependent on whether it is capable of MPLS or not. In particular, MPLS-TE provides the functionality similar to those implemented in ASON in the domains such as discovery, routing, signalling, survivability. The functionality related to QoS depends strongly on the QoS definitions that are different for each Client. 3.5 Client-ASON interworking issues ASON should provide different Clients with the OCh services as defined in [1]. In this section, we study interworking of ASON with its clients from the point of view of the functionality. 3.5.1 Basic considerations on Client-ASON interworking Up to now, the functionality required for the OCh services and for Clients is separately studied. In the following sections, we consider their relationship. 3.5.1.1 Functionality in the Client-ASON relationship The functionality at the Client layers could be categorised as a function of the similarity to the ASON functionality. The Client functionality could be: Different from ASON's; The same as; Between the same and different. The first case refers to the functionality in which related services carried by the Client are different from ASON's ones. Functionality of the second category is providing the same service as one of ASON's services. Table 1, not exhaustive, presents some examples of the ASON functionality and the availability of their counterparts at each Client layer. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 33 (93) ASON functionality Client functionality availability Frame Relay ATM SDH IP IP-MPLS-TE Control plane Y (Layer 3) Y N (N) Y Connection type PVC/SVC VP/VC Phys. - LSP* Signalling In-channel sig. PNNI - - RSVP, CR-LDP Topology discovery N Y: Hello - Y: Hello Y: Hello Routing (CR-based) N CR-based - Desti. CR-based End-to-end protection N Y Y** N Y Y N (Y) Y Distributed reroute restoration Automatic end-to-end provisioning N Y N - Y Policing N Y N (Y) Y CAC N Y N (Y) Y QoS handling N Y N (N/Y) Y Cross-connection (Optical) - - Digital - - Demux/mux (Optical) Statistical - Digital - - Y/N: Yes/No; VP/VC: virtual path/channel; PVC/SVC: permanent virtual circuit/switched virtual circuit; *: connection-like; CR: constraint-based routing; (Y): conditional Yes; Desti: destinationbased; Phys.: physical; PNNI: private network node interface protocol; LSP: label switched path; RSVP: Resource Reservation Protocol; CR-LDP: CR Label Distribution Protocol; Hello: Hello protocol; In-channel sig.: in-channel signalling; Y**: Yes with TMN. Table 1 Some examples of the ASON functionality and their availability in the client layers. 3.5.1.2 Perturbation schemes for the Client-ASON relationship and interworking The overlay model supposes, by definition, that the Client layers are clearly separated from the ASON layer and there should be no interference between them except for the services using UNI signalling interfaces. When interference occur, they could be: ASON perturbs the operation of the Client; A Client perturbs the operation of ASON; Mutual perturbation. The first case is a very important point for the Client. If ASON perturbs frequently and/or strongly the Client operation, such situation cannot be acceptable for the Client. However, the second case may not be catastrophic for the Client if ASON does not perturb the Client in turn. Are there interworking problems at the functionality level ? This question relates to the fact that some functionality is implemented in both the Client and ASON layers. Obviously, functional implementation itself cannot provoke any reaction. But when the same functionality enters simultaneously (or sequentially) in action in both the Client and ASON layers, what will happen ? Answers to this question could be quite ambiguous: they may range from "severe failures" to "nothing happens". This is because the interworking problem is not directly related to the fact that a particular functionality is (or is not) implemented in both the Client and ASON layers. Let us look at the case where ASON troubles its Clients. This is generally related to the fact that: 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 34 (93) EURESCOM Technical Information During a Client process, ASON changes its network configuration; The Client layers may see a logical Client network topology5 that is not the current one. Client starts a procedure, during the period where the network topology of ASON is not completely stabilised. During this period, frequent loss of signal or connection could occur and could even break sessions down. This is a case of the booting up phase of ASON where any request does not work properly at the Client layer level. A similar phenomenon could be observed when ASON is doing the recovery procedure after failure. For instance, when ASON starts an automatic OCh restoration procedure after a failure, during this phase, the Client layers cannot know changes in the network topology of ASON. Hence, if a Client starts, during this phase, a new connection procedure it cannot be achieved properly. Some Client functions related to topology discovery, routing (via database) and signalling could particularly suffer from such situation. They include end-to-end path set-up process such as provisioning, protection, restoration of VC/PV (ATM) or LSPs (MPLS). In order that a Client device can communicate across ASON with other Client devices, ASON should be transparent to any Client signals. For instance, when ASON uses in-band signalling of a SDH frame, the overhead field of the SDH frame, especially the part being used by SDH, should not be modified. Similar requirements are valid for signalling networks. Thus, ASON operation should not interfere with any client functionality. Therefore the following requirements are necessary to avoid troubles in the procedure of different Client functionality. ASON shall be: Transparent to any Client signals; In steady state operation; ASON should very rapidly change their network configuration so that its Clients can not see it. Or ASON should not frequently change their configuration. Compatible with QoS, Policing, CAC of its Clients. ASON should not alter QoS guarantee or differentiation, Policing or CAC of its Client during its transport process. 3.5.2 Functionality implementation issues in the control plane As an overlay model network, ASON should not show its internal resources to its Clients. Since signalling of ASON, through UNI signalling interfaces, provides its Clients with Client-to-Client connection requests across ASON, the Clients do not need to know the entire ASON resources. It can also be case that the network operator providing such services does not want to give its Clients such information. In this context, it should be suitable to separate some signalling functions from the functionality that handles network resources. One possible scheme is to partition the control plane into two entities, for instance higher and lower planes, and to allocate each of them the control plane functionality by taking into account the above considerations. For instance, A higher plane with signalling functionality that can communicate with its Clients through UNI interfaces and A lower plane dedicated to discovery, routing, database management, etc. The higher plane could correspond to the usual one and the lower could be nearby node elements (see Figure 2). 5 The Client layers can not see the network topology of ASON. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information 3.5.3 page 35 (93) Conclusion regarding functionality distribution/modification The functionality in the Client layers is independent of ASON and neither be removed from nor can be moved to. On the other hand, the OCh services provided by ASON determine its functionality. In this context, there would not be any particular implementation issues apart from keeping the control plane functionality as described above. In order to prevent from troubles in the network operation, ASON requires appropriate network solutions that minimises the duration and frequency of transient states. A two-level control plane concept is proposed in order to prevent from undesirable interactions with Clients who do not belong to the administrative domain of ASON operator. 3.6 Conclusion regarding Client layer vs. ASON functionality ASON is based on the overlay model and should provide its different clients with the OCh services defined in [1]. Therefore, ASON should cope with different functionality of its Clients. This chapter presented a description of ASON from the perspective of functionality. The functionality was described from two perspectives: core-edge and client-server. This description allowed identifying functionality to be implemented at different network positions. The core functionality that is mainly associated with switching of optical channels includes topology discovery, optical routing, signalling, policing and CAC. As for those of the edge that deal with various controls of incoming and outgoing traffic streams include QoS handling, connection handling, CAC, policing and UNI interface functionality. These functions cooperate to perform automated end-to-end OCh processes such as provisioning, protection or restoration. Functionality of the UNI interface allows some Clients to request a connection directly. The description of functionality required for handling QoS revealed a lack of definition for OCh services. Nevertheless, QoS handling issues should be addressed at the edge nodes. The client-server description where ASON is seen as the server of the client layers could help understanding how ASON could interact with its Clients and what are main consequences. The server layer functionality can be grouped into the control plane functionality and OTN layer one. The former is very similar to the functionality described from the core-edge perspective and the latter refers to the native functionality of the OTN layer such as optical cross-connection. We have attempted to create a short summary of the Client functionality in order to determine the Client-ASON relationship at the functionality level. We have included Frame Relay, ATM, SDH and IP into our investigation. The results show a strong similarity in the functionality between ATM with PNNI protocol and MPLS-TE based IP and the completeness of their functionality for handling OCh services. Finally, a study on the interworking of ASON with its Client layers suggests that ASON could perturb its Client layers' activities during its reconfiguration phase, for instance, after a failure. However, no particular implementation issues are expected at the functionality level. Table 2 summarises functionality implementation in each node or plane. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 36 (93) EURESCOM Technical Information Functionality Edge Core C-plane OTN Policing x x x CAC x (x) x QoS handling x Signalling x x x Optical routing, Discovery x x x Automated end-to-end processes (provisioning, protection, restoration) x x x Distributed database management (MIB) x x x Grooming x NE level protection x x x Optical cross-connection x x x Failure detection x x x Performance monitoring x x x (x) (x) x x Table 2 ASON functionality implementation EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information 4 Management requirements 4.1 Introduction page 37 (93) As a telecommunication network, ASON should, in principle, support the management capabilities that are common in telecommunication networks. ASON should in addition support both centralised and distributed management as a function of the service provided. In general, management of telecommunication networks covers the following five areas: Performance management; Fault management; Configuration management; Accounting management; Security management. In the context of ASON, however, some requirements for these management areas could be different from, for instance, those defined for the telecommunications management network (TMN) or the optical transport network (OTN) and some others could remain the same. These are determined by whether management uses the control plane or not. The management requirements of OTN have been studied in EURESCOM P918 project [6]. Generic network requirements are defined in G872 [7] for the OTS, OMS and OCH layers of OTN [Annex D]. Most of these management requirements are also applicable to Automatically Switched Optical Network at the OCh layer, as long as ASON is an OTN based on an intelligent control plane implemented in the network elements. In this chapter, first we consider, referring to management of TMN [8], the validity of management approach for each of the five management areas and comment on needs for ASON. We then focus on specific management aspects required for ASON: What are the specific management requirements for the ASON control plane ? What are the differences between distributed management and centralised management ? 4.2 Survey of management requirements In this section the validity of TMN management approach for ASON management in each of the five management areas is being considered. 4.2.1 Performance management Performance management provides functions to evaluate and report upon the behaviour of telecommunication equipment and the effectiveness of the network or network element. Its role is to gather and analyse statistical data for the purpose of monitoring and correcting the behaviour and effectiveness of the network, NEs or other equipment and to help in planning, provisioning, maintenance and the measurement of quality. Performance management includes: Performance Quality Assurance; Performance Monitoring; Performance Control; Performance Analysis. ASON requires Performance management at OCh and Client levels to monitor the conformance of end-to-end OCh service performance to the corresponding SLA. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 38 (93) 4.2.2 EURESCOM Technical Information Fault management Fault management is a set of functions that enables the detection, isolation and correction of abnormal operation of the telecommunication network and its environment. Fault management includes: Reliability, Availability, Survivability (RAS) Quality Assurance; Alarm Surveillance; Fault Localisation; Fault correction; Testing; Trouble Administration. ASON requires Fault management capabilities at OCh level and control channel. 4.2.3 Accounting management ASON requires Accounting management enabling the measurement of the use of end-to-end OCh services and the determination of costs to the service provider and charges to the customer for such use. It also supports the determination of prices for services. 4.2.4 Security management ASON requires Security Management, which includes: Prevention; Detection; Containment and recovery; Security administration. One particular security issue for ASON is related to interfaces (UNI, NNI) with the control plane. At the public UNI, authorisation and authentication should be required. The public interfaces (across the administrative boundaries) and private interfaces (within the administrative boundary) shall have different security level. The public UNI/NNI should have more stringent restriction in terms of routing information exchange, security access control. 4.2.5 Configuration management ASON requires Configuration management that provides functions to exercise control over, identify, collect data from and provide data to NEs. Configuration management is mostly done in a distributed way by using the control plane. It includes: Network planning and Engineering; Installation; Service Planning and Negotiation; Provisioning; Status and Control. Among these five areas of Configuration management, "Provisioning" and "Status and Control" areas are specific to ASON management. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information 4.2.5.1 page 39 (93) Network planning and engineering This deals with the functions associated with determining the need for growth in capacity and introduction of new technologies. It involves evaluation of alternate plans and the entry of chosen plans into a database that will support the Provisioning function group. This can be applied to ASON. 4.2.5.2 Installation The network management system should support the installation of equipment that makes up the telecommunication network. It covers also the extension or reduction of a system. Some NEs call for the initial exchange of data between themselves and the management system. This can be applied to ASON. 4.2.5.3 Service planning and negotiation Service Planning and Negotiation deals with planning for the introduction of new services and with those customer contacts to establish new services, change service features and disconnect services. This can be applied to ASON. 4.2.5.4 Provisioning Provisioning consists of procedures necessary to bring equipment into service, not including installation. Once the unit is ready for service, the supporting programs are initialised via the control channel. The state of the unit, e.g. in-service, out-of-service, standby, reserved, and selected parameters may also be controlled by provisioning functions. In the context of ASON, provisioning mostly refers to end-to-end OCh provisioning and is managed in a distributed way. 4.2.5.5 Status and control ASON should provide the capability to monitor and control certain aspects of the NE on demand. Examples include checking or changing the service state of a NE or one of its sub-parts (in-service, out-of-service, standby) and initiating diagnostics tests within the NE. Status can be stored in MIB. 4.3 Specific management requirements for the ASON control plane In this section, we describe management requirements specifically related to the control plane of ASON that introduces differences in the management property. Configuration management is typically affected by the ASON control plane as it allows to perform automatic operations such as end-to-end optical path set-up or path restoration in case of network failure, usually done differently by a centralised management system. 4.3.1 Topology management 4.3.1.1 Node topology management The node topology can be characterised by its physical ports, its switching matrix, its cards and other equipment information. The node topology is stored in the management information base (MIB) of the node. The node topology management consists of managing the node attributes such as administrative state (up/down) and operational state (enabled/disabled) of ports, switching capabilities (which port can be connected to which one). What is specific to the ASON control plane ? : each node should advertise the adjacent nodes of any change on its current ports and switching attributes (creation, deletion, state change). 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 40 (93) 4.3.1.2 EURESCOM Technical Information OCh link topology management Each OCh link can be characterised by two termination points, which are peer ports of two nodes. The link topology is stored in a distributed database over each node MIB : each node stores in its MIB the link information of the links between itself and its neighbours. The management of OCh link topology consists in managing the link attributes such as the link state, the link capacity (if it is a composite link of sub-links). What is specific to the ASON control plane ? : each node should advertise the adjacent nodes of any link state change. 4.3.1.3 Optical path topology management The path topology is stored in a distributed database : for example, each node that begins a path can store this path in its MIB. The management of the path topology consists in managing the path attributes such path identifier, path state (active/not active), list of nodes and their associated ports which compose the path, etc. What is specific to the ASON control plane ? : in case of an automatic path restoration after a failure, the path attributes should be automatically updated. 4.3.2 Optical path routing By definition, optical path routing is specific to the ASON control plane. Optical path routing is a server process, which calculates optical path from endpoint “A” to endpoint “Z”. This process is part of the ASON control plane : each node executes this process. Optical routing serves bandwidth demand coming from NMS or from UNI. We can define following requirements for optical path routing: fast routing algorithm (for instance, find a route in less than 5 milliseconds from a database of a mesh network of 30 nodes); routing based on constraints; Management of database for optical routing. 4.3.3 End-to-end path provisioning The end-to-end path provisioning consists in three phases: A path request is sent at NMS interface or UNI interface; The route is calculated (done by optical routing); The optical path is set-up. A signalling protocol is required in the ASON control plane to support messages, which will set-up the lightpath into the nodes (set the cross-connections between ingress and egress ports). 4.3.4 Control channel management A signalling network is needed to implement the ASON control plane. A control channel is required between each nodes, and between one (or more) node(s) and the NMS. The control channel between each node can be “in band” (part of OTN signal) or “out of band” (external network). The requirements for the management of the control channel consist in: configuring NMS IP interface address; configuring “in band” or “out-of-band” mode; providing for a control channel failure. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information 4.3.5 Protection management 4.3.5.1 Difference between protection and restoration page 41 (93) Usually, protection means reservation of shared or dedicated network resources to protect a working resource. This is the case in SDH networks with SubNetwork Connection (SNC) protection, and MS-SPRing protection (Multiplex Section Shared Protection Ring), and linear MS protection. 1+1 protection allocates simultaneously working and protecting resources whilst 1:1 and 1:n protection allows extra traffic with low priority on the protecting resource. 1:1 and 1:n protection need an Automatic protection Switching Protocol (APS). Restoration does not need a pre-allocation of protection resources. In case of failure, the restoration process will find automatically a new path among available network resources, and set-up dynamically the new path after releasing of the old one. 4.3.5.2 End-to-end path protection For a very high quality of service, it might be necessary to support within ASON standard path protection schemes such as : 1+1 Path protection 1:1 Path protection 1:n Path protection The ASON control plane should pre-calculate and reserve in advance the protecting path. 4.3.5.3 End-to-end path restoration End-to-end path restoration with standard reroute is basically one of functionality of that the ASON control plane can implement. Because ASON can already make automatically end-to-end path provisioning, end-to-end path restoration should be automatic and fast enough (about 100 milliseconds). 4.3.5.4 Link protection It might be useful that the ASON control plane support 1+1 and 1:1 link protection, against port failures or link failure if the working and protecting links don’t belong to the same fibre. 4.4 Distributed and centralised management In the centralised scheme, network elements (NEs) are directly managed by the network management system (NMS) whereas in the distributed one, they can be managed indirectly or autonomously. The way in what NEs are managed relates closely to the network architecture. In this chapter, we discuss on characteristics of these two management types. 4.4.1 Distributed management Figure 5 describes the distributed management architecture of ASON. The network MIB is distributed among the nodes. Each node has a partial knowledge of the global network MIB: it knows the network attributes related to adjacent link and node topology (adjacent links and nodes), and a subset of the path topology. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 42 (93) EURESCOM Technical Information Figure 5 Distributed management (ASON) Additional network attributes, which are generally administrative attributes set by NMS operator, are stored in the NMS MIB only. (This can be usernames for paths, comments, etc) The equipment MIB is stored in each node and can be copied in the NMS MIB for MIB download in case of MIB recovery after a crash of the node (but this is not mandatory: it depends on the NMS and the equipment). 4.4.2 Centralised management Figure 6 describes the centralised management architecture. The network Management Information Base (MIB) is centralised and maintained only in the NMS MIB. The equipment MIB is stored in each node and can be copied in the NMS MIB for MIB download in case of MIB recovery after a crash of the node (but this is not mandatory: it depends on the NMS and the equipment). The management architecture is based on the agent/manager model. Management protocol used at NMS interface could be CMIP or SNMP protocol. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 43 (93) Figure 6 Centralised management (OTN) 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 44 (93) 4.4.3 EURESCOM Technical Information Comparison between distributed management and centralised management (Table 3) Distributed management Centralised management Client/distributed Server model Agent/Manager model Control plane = Server NMS = Client Node = Agent NMS = Manager NMS MIB includes image of distributed Network MIB and image of Equipment MIB includes network attributes (unique centralised database) and image of Equipment MIB MIB inside Node includes Equipment attributes and Network attributes (distributed network database) includes only Equipment attributes. Network and Nodes State handling handled automatically within the control plane handled by NMS software upon state change notifications received from agents. Optical path routing Background Routing protocol task executed by each Node software. Manual Path computation or automatic path computation done by NMS software. Management model Automatic path computation No Network level information. Optical path set-up Done by Nodes within the control plane. Done by NMS requests. Optical path restoration Done automatically by Nodes within the control plane. Can be done in less than 100 ms. Done by NMS software. Generally, it needs more time (several seconds or minutes) due to NMS interface delay. Notifications between Nodes and NMS Automatic update of NMS MIB in case of state change or path restoration Automatic update Table 3 Comparison between distributed and centralised management 4.4.4 Advantages and disadvantages 4.4.4.1 Centralised management Disadvantages: Complexity: Centralised management systems for transport networks use complex information models (for example G.774 for SDH networks) which have been defined in G.805 and M.3100 ITU-T recommendations, and which follow an object oriented modelling approach: physical network resources (links, ports, …) and logical transmission resources (trails, trail termination point, connection termination points, …) are modelled by objects. Object instances follow specific naming rules, and are stored in the centralised NMS MIB. Information model for protection schemes are not always standardised and are generally complex to implement; MIB update latency: due to NMS interface delay; Not “real-time” operations; Advantages: EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 45 (93) Unique network MIB to maintain: no need of MIB synchronisation with the nodes. MIB upgrade not difficult when the MIB format change: MIB upgrade doesn’t impact the nodes. 4.4.4.2 Distributed management Disadvantages: MIB upgrade more difficult when MIB format changes: the MIB format should probably be compatible with the previous one. MIB upgrade should be done in every node without traffic interruption; Need more robustness and reliability of the node software, especially the routing and signalling software components. Advantages: 4.5 Automatic end-to-end path provisioning; Automatic and fast path restoration in case of a network failure; Allows future interoperability with client equipment at UNI interface, or future integration with client management system (peer network model). Summary Starting from the general management requirements, the specific management requirements for the ASON control plane have been described. These include the management of the topology (i.e. node, link and path), end-to-end paths, control channel, protection/restoration and UNI interfaces. Also a comparison is provided of the distributed management approach of ASON and that of conventional centralised management, such as TMN. The advantages and disadvantages of each management approaches are summarised together with their architecture schemes. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 46 (93) 5 EURESCOM Technical Information Grooming issues ASON can handle the automatic switching of big pipes, i.e. optical channels, but it cannot deal with a smaller granularity than the OCh. Therefore the end nodes of ASON must be associated with a grooming functionality that can put in the same OCh different clients of smaller granularity. This grooming function is as important as ASON itself because this function that corresponds to the ASON-client interfaces performs the filling of the OChs. The automatic access, the protection and the fast provisioning of small granularity channels (from 100 Mbit/s up), which are groomed in one OCh, are conditions for the future success of ASON. However, e.g. at the ITU-T, several vendors are proposing ASON-like networks with a granularity smaller than the OCh on the bases of their commercial products. These solutions correspond to ASON-SDH networks where grooming can be performed both in the core and at the edge nodes. 5.1 Definition Aggregate Add together several client signals to one single signal. A general term. Grooming The allocation of server layer trails to client layer connections which group together client layer connections whose characteristics (such as service type, destination or protection category) are similar or related (from ITU-T G.783). This term is mostly used for source-to destination aggregation. Multiplexing Combining several signals over a single path through TDM, WDM,… Often used for hierarchical systems, such as SDH. In a network multiplexing assumes a preceding grooming of the individual connections to minimise the number of multiplexing operations. 5.2 Requirements for grooming function The grooming function has to answer the following question: How can we use the ASON functionality, which is attributed to high granularity OChs (2.5 Gb/s, 10 Gb/s) for clients with granularity of 100 Mb/s to 1Gb/s? Grooming is performed by the client-ASON interface placed at the edge ASON nodes (and in the core if an ASON-SDH network is used) and should be performed taking into account some traffic parameters, like QoS, signal type, etc. This interface must have the knowledge of the used and unused ASON OChs and of the filling degree of each OCh. By this way, it can be decided whether a new OCh has to be created or if a partly filled one can be used. The grooming function is responsible for the best use of the ASON resources. In the first phase, this functionality is not very complicated to implement. The minimal number of OChs has to be used between nodes A and E. Only in case of congestion (all OChs are filled), one new OCh between nodes A and E is opened. However, a real network is more complicated. For nodes that are situated far away from each other, and that do not have a high traffic, the creation of an OCh is too costly. One wavelength would be reserved for the whole length. This traffic has to be groomed with other traffic between nodes of smaller distances, as shown in Figure 7. A B C A B __________________C D E D E Figure 7 Clients between two distant nodes should be groomed several times with other clients in order to use the minimal number of OChs. Without grooming an extra OCh first in red. With a grooming in C and D already opened OCh are used. An algorithm must be used to define in what kind of condition, the client has to be groomed or not. One possible algorithm would be to use the Open Shortest Path First (OSPF) routing algorithm to EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 47 (93) choose a route through the node and fibre infrastructure without taking account of the wavelength. Along the chosen route, a connection that uses the minimum number of wavelengths should be selected among the possible connections composed of OCh segments (or wavelengths) having an enough free capacity. To do this, the longest segment of already opened OCh that has enough free capacity must be used. Here the longest segment means a segment of a single and same wavelength that traverses the maximum number of nodes. Between each OCh a grooming function must be used. Ideally, the same functionality should be attributed to the channels that are groomed as the functionality of ASON optical channels: configuration, fast provisioning, protection. All these functionality must cooperate with the grooming function in relation with ASON. 5.3 Grooming and services in ASON Regarding the different services to be provided by ASON, grooming the client signals at the ASON edge can be taken into account to determine specific services adapted to the client signal. Several options in ASON architecture contribute to the implementation of such functionality. 5.3.1 Option: ASON-SDH Client A Client B A service Signal type QoS requirement … B service Signal type QoS requirement … OXC grooming OCh ASON Control plane OXC routing OXC operates grooming ASON control plane takes into account client signal OXC Figure 8 Grooming the client and client service requirements are taken into account by ASON: this is the ASON-SDH option. Client services can be defined and taken into account within ASON. ASON can handle a small granularity. The routing and resource allocation process can be done at the client signal granularity. In this scenario, optical cross-connects may be able to handle the various client interfaces. Moreover, routes can be determined and established at a lower granularity than the optical channel (VC-4 for instance). This scenario corresponding to the ASON-SDH network is illustrated in Figure 8. 5.3.2 Option: ASON-OTN Another option is to map the client services into ASON optical channel level services. ASON can only handle the optical channel granularity. This option simplifies the protocols used for ASON control plane, at the expense of a constraint in the grooming process (only client services of a given type are multiplexed in one optical channel) 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 48 (93) EURESCOM Technical Information and minimises the expected gain in resource. This option corresponds to the standard ASON or ASON-OTN with grooming at the edge and is illustrated in Figure 9. Client A A service Signal type QoS requirement … Client B B service Signal type QoS requirement … Multiplexing equipment Grooming OCh service OCh Mapping client service QoS Into OCh service type OXC OCh ASON Control plane OXC routing Multiplexing equipment operates grooming OXC Figure 9 Grooming is managed at the ASON edge. This is the standard ASON or the ASON-OTN. 5.4 Grooming scenario The grooming can be performed by different protocols: IP, ATM, SDH. Due to the suppliers' solutions and the ITU-T standardisation process, the actual best grooming scenario is to use SDH. However, as the IP clients are already groomed by the routers, they do not need to use the SDH grooming. The IP transport can be done by using POS or GbE interfaces. As shown in Figure 10, G.709 (or digital wrapper) can be used to groom between 2.5 and 10 Gbit/s. IP router (possible POS) or GbE switch/router STM1 ATM FR GbE SDH client interfaces G.709 Multiplexer lambda channels Multiplexer Figure 10 Proposed grooming scenario, by using the SDH. The router and the SDH multiplexer can also have a direct access to a wavelength (in red and blue). G.709 is only shown as one possibility. In this scenario the client interfaces are client-SDH interfaces. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 49 (93) Before the SDH multiplexer, client mapping interfacing are necessary for ATM, Ethernet, GbE, FR etc. One very strong reduction of the ASON flexibility corresponds to the client interfaces, which must be present in the edge. No new client channel can be opened without their presence. To gain in flexibility reserved Ethernet, ATM or GbE interfaces must be implemented. 5.5 Summary Grooming is necessary to handle sub-OCh rate traffic streams. It improves the granularity and flexibility of services of ASON by using efficiently OCh connections. Two grooming schemes (ASON-SDH and ASON-OTN) are considered at the ASON edge. ASONSDH refers to the case where both core and edge nodes have grooming capabilities thanks to SDH multiplexing functions. Such ASON could directly deal with sub-OCh rate streams. Whereas in the second one, ASON handles only OCh rate traffic streams and the incoming client traffic streams are groomed at the edge. Finally, we propose a grooming scenario based on the ASON-OTN scheme. In this scenario, the G.709 digital wrapper is used. SDH multiplexers and IP routers with POS or (10)GbE interfaces provide additional possibility for grooming. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 50 (93) 6 EURESCOM Technical Information Optical transparency related issues ASON should provide customers with several new services thanks to its automatic functions implemented in its control plane. The basic functionality of ASON is defined for providing these services and described in Chapter 2. Since ASON can be either opaque or transparent or a mixture of both, the services to be provided by ASON should be independent of transparency issues of ASON. Hence both opaque and transparent ASONs require a set of functions that cope with these services. Starting from the definition of optical transparency, we describe in this chapter impacts of optical transparency upon the functionality and technological issues related to the implementation of the signalling network and network elements. 6.1 Definition of optical transparency In general, we can say an optical network is transparent when any characteristics, including the format, the data rate, protocols, of the original signal travelling over the network does not suffer from any modification and the signal preserves the same quality along the entire transfer path. In particular, in the context of ASON, transparency can be defined from transparency of OChs. An OCh is transparent when: Optical signal carried over the OCh is not subjected O/E (or E/O) conversion in any network elements traversed; The quality of the optical signal carried over the OCh does not depend upon neither the signal format nor bit rate (optical signal carried over the OCh can be considered as the same with respect to signal characteristics). A transparent OCh is practically defined by the absence of O/E (or E/O) conversion within the OCh. An optical network is transparent when any end-to-end OCh belonging to the network is transparent. 6.2 Impacts of functionality optical transparency upon the ASON's ASON should provide a set of functions for providing OCh services described in Section 1.1. This must be independent of optical transparency issues. 6.2.1 Functionality implementation in transparent network elements As defined above, a transparent ASON does not have any network element subjected O/E (or E/O) conversion. This can provide not only some benefits but also some drawbacks. Hereafter, referring to some key network elements, we describe how transparency could influence on functionality implementation in transparent network elements. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information 6.2.1.1 page 51 (93) Transparent optical cross-connect node The basic function is to space-cross-connect (SXC) the input ports to the output ports. The node has not any O/E or E/O conversion elements. The incoming signals to the ports: 1,2 and 3 of OXC 1 that stem from different points of the network are going out from the ports: 4,5, and 6 of OXC 2 to different destinations. (1) Each port is connected to an own fibre carrying a wavelength (or a set of wavelengths): OXC 1 OXC 2 IN 1 1 2 3 5 6 3 5 1 4 2 1 2 3 OUT 2 6 2 1 4 3 3 Figure 11 Multi-fibre interconnection scheme. Fibres are space-cross-connected and wavelength cannot intervene in the port interconnection relationship of two adjacent nodes (Figure 11). Wavelengths go through OXCs and remain the same. Wavelengths carried by a particular fibre are not known. Only the basic function that is space cross-connection of fibres is performed. (2) Ports are optically multiplexed: 2 OXC 1 1 OXC 2 OMS 1 6 1 2 2 5 2 3 3 4 3 DEMUX 1 1 2 3 2 5 3 2 4 3 MUX 6 1 DEMUX 1 MUX Figure 12 WDM interconnection scheme. An OMS section is inserted in-between (Figure 12). Since each port of optical multiplexer (MUX) (or demultiplexer (DUMUX)) corresponds to a particular wavelength, wavelength conversion that shifts, for instance, from 2 (3) to 1 (2) at port 6 (5) of OXC 1 is required at the output ports of an OXC. Wavelength ensures an appropriate interconnection of ports of two adjacent nodes. To provide the space switching function, an OXC requires wavelength conversion. Wavelength cannot be continuous between two adjacent nodes. This configuration requires MUXs, DEMUXs and fully optical wavelength converters (mandatory). Note that if wavelength is a set of wavelengths centred at i and the wavelength window of the multiplexer is large enough to passing through these wavelengths, space switching is done on a set of wavelengths. 6.2.1.2 Optical transparent links There is no O/E or E/O conversion in any element of the OMS/OTS sections. In particular, this requires the use of optical MUX, DEMUX and fully optical regenerators. Wavelength has to remain continuous over the section. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 52 (93) 6.2.1.3 EURESCOM Technical Information Functionality implementation issues for transparent network elements In a transparent OXC, where fibres are space-cross-connected, incoming wavelengths go through and arrive at the egress ports without any conversion. To support a function other than space-crossconnection, the OXC requires additional equipment, for instance, monitoring equipment to monitor wavelength that are not present and of which network solutions have to be found. Their implementations in general could be more difficult in a transparent OXC than in an opaque one. Without transceiver (TX/RX) at each port, in-band signalling is not available. The functionality such as neighbour discovery, port discovery or performance monitoring that uses in-band channel cannot work. Appropriate network solutions are again required to support that functionality. Inserting OMS sections, in which intervene multiplexing and demultiplexing, transparent OXCs require a new functions such as wavelength conversion that completes a basic function, namely optical cross-connection. Therefore, functionality issues have to be considered taking into account physical implementation of the OMS/OTS section (e.g. some aspects of the optical signal transport). Whereas, in an opaque ASON, characteristics of OMS/OTS section are independent of ASON operation. The design of OMS/OTS section can be considered as engineering of transport links. To support all the basic functionality (that refers to the minimum and sufficient functionality to cover the defined services), transparent network elements require appropriate network solutions as well as some additional functions. The former stems from the fact that the current signal processing systems are based on electronic signals instead of optical signals and O/E conversion is required before any signal processing. This evidence promotes opaque network elements rather than transparent ones. 6.2.2 Functionality issues in a transparent ASON 6.2.2.1 Degree of influence upon the basic functionality ASON can be based on either opaque or transparent nodes. As described above, a transparent node requires additional functionality to support a basic functionality. In addition, some functions could not work at all without appropriate network solutions. Let us now look into functionality issues in a transparent ASON. The basic functionality could be differently affected and be categorised as follows: unaffected basic functionality; The functionality is independent of transparency issue. CAC could be an example. basic functionality requiring some appropriate network solutions; This functionality can be implemented as is if appropriate network solutions are implemented. Examples could be functions involving monitoring functions such as fault detection, neighbour discovery, traffic policing, etc. basic functionality requiring appropriate database; Some functions require a database that takes into account transparency and resulting new functions. An example is routing that requires a database including wavelength conversion table, path length, etc. basic functionality requiring additional functionality; Some functions require additional functions that work together. An example is optical space-cross-connection function over WDM that requires wavelength conversion function. additional functionality. This functionality does not belong to the basic functionality and works in association with a basic functionality. An example is wavelength conversion function in the case of optical space-cross-connection function over WDM. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information 6.2.2.2 page 53 (93) Basic functionality issues Neighbour (port) discovery: To perform neighbour discovery, messages emitted from a port of the local node over a fibre (data channel) have to be received by a neighbour node (remote). Obviously it requires signal-monitoring function achieved by a transceiver at each port. In an opaque node, it is easy to implement a transceiver at a port whereas it is not trivial in a transparent node. Routing: Routing requires a database describing the state of topology (or links in the link-state routing) in which include inter-nodes transmission characteristics (fibre length, bit-errorrate, optical gain/loss, noise, …, possibly the use of optical performance models and parameters) and wavelength converter characteristics in addition to standard information such as port characteristics (bandwidth, metric, …). Transparency can increase the amount of data to be handled. Algorithm for path computation can be more complex than standard one (for instance SPF). Signalling: Signalling in ASON provides functions, like set-up, maintain and release of an optical path. Signalling messages are sent over the signalling network that can be in-band or out-of-band or a combination of both. To provide in-band signalling, a transparent ASON obviously requires network solutions coping with signalling message transmission and reception over fibre. Protection/restoration (OCh level survivability): OCh level protection/restoration involves several functions: fault detection, signalling, routing, etc. Clearly, to provide this functionality a transparent ASON requires a set of different network solutions and specific components related to transparency. 6.2.2.3 End-to-end functionality issues An end-to-end functionality refers here to the functions performing handling of an end-to-end path and is built on the basic functionality. Some examples are automatic end-to-end path provisioning, and end-to-end path protection/restoration. Handling automatically end-to-end paths in a transparent ASON involves mainly routing and signalling functionality and then requires a set of different network solutions and specific components related to transparency. 6.2.2.4 Advantages and drawbacks in the functionality In the case of OXC, the lack of O/E/O conversion makes the function of space-cross-connection independent of the bit-rate, signal format, protocols etc. This advantage obtained at the data transport level can be counterbalanced by the difficulty in implementing other functions than space-cross-connection still being required. In particular, it is actually difficult to get network solutions for monitoring functions that are involved in neighbour discovery, fault detection, traffic policing, performance monitoring, etc. Then there are functions that could not work at all without appropriate network solutions. After the insertion of MUX or DEMUX between two transparent nodes, space-cross-connection cannot work without wavelength conversion. Thus there are functions that require new functions completing them. From the point of view of the functionality, a fully transparent ASON could show more drawbacks than advantages. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 54 (93) 6.2.3 EURESCOM Technical Information Management issues The impacts of optical transparency are seen from the point of view of the management. 6.2.3.1 Node management A node is characterised by its physical ports, its switching matrix, its cards and other equipment information. In the case of an opaque node, the knowledge about wavelength is not mandatory for the management whereas it is mandatory in a transparent node. Therefore, transparency could increase the information to be managed. 6.2.3.2 Link topology management In addition to the link-state database (describing the state of interfaces), the management system requires the information about characteristics of OMS/OTS sections. Transparency could modify the contents of the link topology database. 6.2.3.3 Path topology management The management of the path topology should have to take into account the path attributes resulting from optical transparency. 6.2.3.4 Performance management Today, monitoring the optical signal in a transparent optical network remain one of the biggest challenges; hence, management, performance and multi-vendor interoperability still constitute an obstacle to the deployment of transparent optical network. Several elements could be mentioned. First, the recent advances in all-optical functions towards all-optical wavelength conversion and all-optical regeneration are very promising, however the industrialisation of such key components for the transparent approach still seems to be beyond the introduction of the first optical crossconnecting function in the optical layer. Another element is the possibility of “transparent” monitoring of the signal quality. 6.2.4 Conclusion Optical transparency could extend the capabilities of a node. Since the basic functionality required for providing the OCh services must be independent of optical transparency issues, functionality issues in a fully transparent ASON consist in finding appropriate network solutions, implementing additional functions and modifying appropriately its databases. 6.3 OCh length, transparent islands, resource allocation 6.3.1 OCh length in a transparent ASON 6.3.1.1 OCh length issues In a transparent ASON, as pointed out above, the signal quality at the optical transport layer has to be taken into account for routing, in particular for determining end-to-end OCh paths. In general, an acceptable signal quality is ensured between 3R points within a given network engineering rule. And the maximum distance between them determines the maximum length of an OCh. Therefore when determining automatically a working OCh and a protection or restoration OCh, the length of those OCh has to be shorter than the maximum length. In addition, it can also determine the maximum dimensions of a transparent island within an optical network and resource allocation strategy. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information 6.3.1.2 page 55 (93) Engineering limitations Considering today’s technology, the interest for transparent optical nodes appears to be linked to the engineering limitations, that is, the maximum length of the optical span. Thus the association of ultra-long haul systems (using soliton technology and/or Raman amplification to increase the maximum length of the optical path allowed without any OEO regeneration) and optically transparent cross-connects (supposed to be cheap due to the interface transponder savings) is probably the most relevant scenario. 6.3.1.3 Maximum transmission distances The maximum length of an OCh depends on several physical factors that induce impairments of optical signals, such as reductions in the optical signal to noise ratio, or chromatic and polarisationmode dispersion, or fibre non-linearities which can accumulate from span to span. However, we will focus on the number of wavelengths, bit rate and fibre type in the following. If we consider a single OCh with 2.5Gb/s in one fibre, the maximum distance is nearly unlimited, since we can optimise and compensate the dispersion without need to compensate the dispersion slope. Several thousands of kilometres can be reached before amplifier noise dominates. Even for a single OCh with 40Gb/s it’s possible to reach 500km [9]. As soon as several DWDM channels are considered, new physical effects arise, which significantly reduce the maximum transmission distance. On one hand the dispersion compensation cannot optimally be done since the dispersion slope in fibres isn’t equal. Non-zero-dispersion fibres work well for a small wavelength range. But if 100 wavelengths dispersed between 1530nm and 1560nm are carried, even non-zero-dispersion fibres could show dispersion slope problems (and therefore dispersion compensation problems). Today several new methods are proposed to mitigate the dispersion slope problem (e.g. Bragggratings for each wavelength) but they imply big expenses. Thus, for a transparent optical island the path length for 2.5Gb/s channels and different number of channels is shown in Table 4. Although the results in this table are from 1998, it gives a good overview on what is possible to transmit on commercial available DWDM systems today. Increase in the transmission distance could be expected by the use of Raman amplifiers in the span. Raman amplifiers can significantly increase the span length. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 56 (93) EURESCOM Technical Information Total carried bit-rate [Gb/s] Distance [km] 2.5 10 20 40 WDM SMF + DCF SMF 100 Single Single WDM 4x2.5Gb/s SMF + DCF or WDM 2x10Gb/s 8x2.5Gb/s 4x10Gb/s DSF + DCF or or 16x2.5Gb/s SMF + DCF WDM SMF + DCF SMF 500 Single Single WDM 4x2.5Gb/s SMF + DCF or WDM 2x10Gb/s 8x2.5Gb/s 4x10Gb/s DSF + DCF or or 16x2.5Gb/s SMF + DCF WDM SMF + DCF SMF 1000 Single Single WDM 4x2.5Gb/s SMF + DCF or WDM 2x10Gb/s 8x2.5Gb/s 4x10Gb/s DSF + DCF or or 16x2.5Gb/s SMF + DCF WDM DSF + DCF SMF 2000 Single Single WDM 4x2.5Gb/s DSF + DCF DSF 4000 Single Single WDM 4x2.5Gb/s SMF + DCF or WDM 2x10Gb/s 8x2.5Gb/s SMF + DCF or WDM 2x10Gb/s 8x2.5Gb/s 4x10Gb/s DSF + DCF or or 16x2.5Gb/s SMF + DCF WDM 4x10Gb/s or DSF + DCF Table 4 Comparison of different transmission techniques versus bit-rate and transmission distance, with Erbium doped fibre amplifiers [10]. SMF+DCF stands for a link comprising standard single mode fibre and dispersion compensated fibre. 6.3.2 Protection/restoration path length issues The protection path is usually longer than the working path. When determining automatically a working OCh and a protection or restoration OCh, the length of those OCh’s has to be shorter than the maximum possible transparent length. This implies some additional constraints for the computation. Some problems related to protection routes calculation are presented in the following sections. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information 6.3.2.1 page 57 (93) 1+1, M:N protection If a Client asks for a connection with a 1+1 protection then the characteristics of the nodes (transparent or opaque) do influence the evaluation of the two routes (working and protection). It is well known that in order to provide a true 1+1 protection, working and protection route have to be as much as possible disjointed. In particular, this means that not only different couples of fibres have to be used but they should also be located in different conduits. Fibres and conduits diversification is already a quite hard task to achieve in the existing networks, in the OTN/ASON could become even worse. If we look at an SDH network, where each section performs a 3R regeneration, the constraints on the route in terms of total length of the spans and total number of nodes traversed are not really a big issue: as long as a complete route diversity between a working and a protection path is needed, it does not matter if the working path is not optimised6. In other words, calculation of working and protection can be done separately. Let’s consider now the ASON/OTN. In case of opaque nodes no big differences are envisaged with the usual (SDH, for example) situation. On the other hand, considering transparent nodes, analogue impairments could put severe constraints on the two routes in terms of the total length and the total number of nodes interested. The working path could occupy resources that could make impossible to find a physically separated protection route. The analogue nature of the signals could lead to the conclusion that, in spite of network resources’ availability, a disjoint protection path can not be set up. The previous considerations have been done considering a 1+1 protection. Generalisation to a M:N protection schemes is quite straightforward, even if it should be taken into account that in a M:N scheme there is no fixed association between a working and a protection path: each of the M protection route should be able to guarantee the fulfilment of the transmission requirements (in terms of analogue impairments tolerated) of each of the N working routes. Of course, this introduces a further degree of complication. From the arguments discussed, the following conclusion can be drawn: ASON control plane should calculate simultaneously the working(s) and protection(s) routes. 6.3.2.2 Restoration Restoration is the most likely to be employed by ASON/OTN because of its potential in reducing the amount of network resources needed to guarantee an high resilience to the offered services. Concerning route calculation, more complications are foreseen for this protection scheme. The principle is always the same: when restoring failed paths, it could easily happen that network resources consumption will soon lead to a situation where no more restoration can occur in spite of plenty of network resources free and available. The differences with protection schemes such as 1+1 or M:N is that pure restoration does not rely on any kind of fixed relationship between the working routes and their protection counterparts: engineering a pure restoration in order to guarantee an optimum (for the specific problem considered) network resources consumption does not look an easy task unless a combination of restoration and pre-calculated (shared) protection routes is considered. In this case the pre-calculated shared routes would be the first to be tried under fault conditions and they could be engineered in such a way to guarantee a reasonable degree of optimisation, leaving the application of pure restoration to the most extreme situations. 6 This is not completely true: route properties (total length mainly) have an impact on the performances of the client signals. In SDH for example, performance objectives for the well-known error performance parameters BBER, ESR, SESR have to be calculated considering the length of the path under consideration. What is important to observe is that in a 3R-nodes based network reasonable (few hundred kilometers) differences between working and protection routes properties have not dramatic impacts on path performances. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 58 (93) EURESCOM Technical Information 6.3.3 Transparent islands 6.3.3.1 Overcoming OCh length limitation It appears that the maximum OCh length is limited by the maximum transmission distance that is determined by signal impairments due to several physical effects occurring during the propagation in the fiber. This problem could be avoided by introducing: opaque cross-connects, performing regeneration at every cross-connect; or optical transparent islands; or hybrid nodes. 6.3.3.2 Optical transparent islands Optical transparent island solution consists of a network with optical transparent and opaque islands. To circumvent the OCh length limitation, the size of transparent islands should not be greater than a size where any OCh length does not exceed the maximum transmission distance. The problem of the signal quality control can be solved by controlling the power along the channel (use of pilot tones: cheap or optical spectrum analyser: expensive) and check the digital quality at the end of the channel (BER). The problem of power equalisation could also be solved by the use of power equalisers, which should in turn increase the costs. Therefore it is important to find a good trade-off between the size of the island and investment. Simulation results have shown that 8 channels at 2.5Gbit/s can be transmitted in networks with a path length of 1000km, having either OXCs or ADMs located every 100km. If we consider that the path length in an optical transparent island must be shorter than the theoretical maximum for security reasons, we can in conclusion say that an optical island with path length 500km and a DWDM System with 8 channels could be a realistic option. 6.3.3.3 Hybrid nodes A hybrid node consists of transparent and opaque parts. It allows using a node as transparent or opaque node as a function of the needs for OCh characteristics. It could be a good alternative to the precedent solutions. However, a network with hybrid nodes could be more complicate to manage (require many algorithms such as: wavelength assignment, algorithm that decide whether to use an opaque or transparent port in a hybrid node). 6.3.4 Resource allocation and wavelength blocking without wavelength conversion 6.3.4.1 Resource allocation The impact of the absence of wavelength conversion on the resource allocation has been studied in the last few years as the “wavelength allocation” problem. The basic idea is that without any wavelength conversion, more resource are required in the optical layer. Theoretical studies based on graph theory have demonstrated that the difference can be low for “static” networks (where demands are known in advance, and no restoration is considered). Other further studies have pointed out the interest for wavelength conversion when dynamic arrival of demands and survivability are modelled. In other words, one could expect that a transparent ASON (or transparent sub-network) would require more optical resource than an opaque one. 6.3.4.2 Wavelength blocking In case of fully transparent network without wavelength conversion, an OCh is carried by a single wavelength, that is, the same wavelength has to be assigned on all links along the OCh from the source to the destination. Hence it is possible for an OCh request to be blocked although there is a wavelength available on every link along the OCh. This is called wavelength blocking which can occur due to the wavelength continuity constraint. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information 6.4 Signalling network implementation 6.4.1 Signalling network implementation alternatives page 59 (93) In principle, three types of signalling can be defined: In-band/in-fibre: the signalling channel is somehow embedded in the optical channel; Out-of-band/in-fibre: the signalling channel is associated with a dedicated optical carrier in the same fibre used to carry optical channels; Out-of-band/out-of-fibre: the signalling channel uses a link physically separated from the link used to carry the optical channels. The applicability of each alternative to a physical realisation of an Automatic Switched Optical Network needs to be studied in association with the characteristics of the ASON’s nodes: opaque or transparent. Before proceeding on, a general assumption is applied in the rest of the document: processing of the information content of the signalling channel can be done only electronically. Whenever the signalling information has to be used in a node, an O/E/O conversion of the signalling channel of interest has to be performed at that node. 6.4.2 In-band/In-fibre This solution can be realised in two ways: either by using a sub-carrier modulation process or by considering a digital wrapper technique. 6.4.2.1 Sub-carrier modulation [11] The signalling information modulates a sub-carrier (operating at a properly chosen frequency), which is superimposed on the current bias of a laser diode in an optical sender which intensity modulates the optical signal operating at a specific λ. This solution has some drawbacks, such as: limited bandwidth available for the signalling channel; no digital performance monitoring of Client signals can be applied; reduced alarm supervision and propagation; scalability problems; implementation of a non-associated shared signalling channel could not be supported. Processing of the signalling channel does not require 3R conversion of the Client signals, so opaque or transparent nodes are both feasible. 6.4.2.2 Digital wrapper This solution assumes that 3R points are always available at the interconnection of different administrative domains: Client signals are supposed to be available as electrical signals at the first OTN/ASON equipment they find and they are logically mapped into a digital wrapper which is a logical signal at electrical frequencies. The digital wrapper defines a frame format and provides bandwidth in its associated overhead for OA&M functionality; it also allows the utilisation of Forward Error Correction codes. This electrical signal then modulates an optical carrier at a proper frequency, the resulting optical signal plus (eventually) a non-associated overhead carried by the Optical Supervisory Channel forms the Optical Channel. Bandwidth could be reserved in the digital wrapper associate overhead to provide transport of a signalling channel. This solution, as well as the previous one, has some drawbacks, for example: 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 60 (93) EURESCOM Technical Information definition of a signalling channel embedded to the associated overhead of the Optical Channel itself could lead to scaling problems; implementation of a non-associated shared signalling channel could not be supported; access to and processing of the signalling channel imply that a 3R point with O/E/O conversion of the Optical Channel is needed. Considering that: OTN/ASON should not require the addition of regeneration functions at points other than those required for transmission impairment mitigation, optical sections’ growth implies that 3R points (and consequently E/O/E conversions) will reduce in number and/or change their locations, ASON should be designed without any need to access OCh associated overhead, otherwise it gets impossible (or, at least, extremely difficult) to remove the functions needed to access that overhead even when the original need for regeneration no longer exist. Being the signalling channel embedded in the digital wrapper frame, an opaque node is needed. 6.4.3 Out-of-band/In-fibre This solution uses a dedicated λ for transport of the signalling channel, for example the Optical Supervisory Channel (OSC) could support this functionality by reserving the required bandwidth in a time slot of the digital signal it carries. With this approach any kind of node, both transparent and opaque, would be feasible. No limit is foreseen in terms of: available bandwidth; scalability. Considering that the OSC is terminated at each optical span anyway and its information content processed, unnecessary 3R points would not be introduced in the network. Moreover, future removal of 3R points (and the associated O/E/O conversions) could be achieved without any impact on the access points to the information content of the signalling channel. Out-Of-Band/In-Fibre signalling is supported either by opaque or transparent nodes. 6.4.4 Out-of-band/Out-of-fibre This solution allows the maximum flexibility concerning the implementation of the signalling channel: virtually any type of physical layer (PDH, SDH, Ethernet) would be feasible. The major drawback is represented by the necessity to build, manage and supervise three networks: OTN (fibres and equipment), ASON (equipment), signalling network. Any type of node would be able to support this type of signalling. 6.4.5 Conclusion Table 5 summarises the discussion on compatibility between node properties (transparency vs. opaqueness) and signalling implementation. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 61 (93) Signalling implementation Supported node type Transparent Opaque Sub-Carrier Modulation Y Y Digital Wrapper N Y Out-Of-Band/ In-Fiber Y Y Out-Of-Band/ Out-Of-Fiber Y Y In-Band/ In-Fiber Table 5 signalling implementation and node properties A signalling based on Out-Of-Band/Out-Of-Fibre and/or Out-Of-Band/In-Fibre solutions are the most likely to be adopted: OOB/OOF in the very first deployment of ASON/OTN because of its simplicity and (maybe) OOB/IF in the future. Both solutions do not put any constraint on the characteristics of the nodes, which can be either opaque or transparent. The choice between them is therefore to be taken considering other aspects (cost, simplicity, management, etc.). 6.5 Wavelength conversion and OXC's properties In order to provide an optimal network resources utilisation, one of the major requirements for the ASON/OTN is the possibility to change the central frequency of the optical carrier associated with an Optical Channel. This capability is usually referred to as “ conversion”. conversion is a feature that has to be supported in the network with the highest priority, in order: to provide a reasonable degree of scalability for the network and properly support dynamic allocation of Optical Channels for automatic set-up/tear down of connections by a Customer; to simplify network planning: having no means to provide conversion would also put severe constraints on the “ consumption” inside the network; to provide efficient re-routing under faulty conditions if survivability schemes based on protection resources sharing, such as restoration, are used: having a fixed association between a and an Optical Channel would rapidly lead to “wavelength blocking” problems. to minimise network expenses: in case of a protection scheme based on restoration and conversion functionality is not supported, the only way to achieve a reasonable degree of reliability would be to provide a great amount of spare link capacity in the network. In spite of recent advances on conversion entirely in the optical domain, this process is still usually carried on through an optical/electrical/optical conversion: if an incoming optical signal S 1 associated to 1 has to be converted to an outgoing optical signal S2 associated to 2, then S1 has first to be converted into an electrical signal and this signal is used to modulate an optical carrier 2 in order to provide the optical signal S2. In other words, conversion also implies the need/possibility to have 3R points. This document assumes that conversion is performed by an O/E/O conversion (except for one case where both alternatives are investigated). Once that it is agreed on the importance of the ability to change the associated with an Optical Channel, it can be faced the problem on how to provide this capability. In order to have a better understanding of the problem, a network scenario is fixed: based on this scenario, three different alternatives for conversion implementation are presented. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 62 (93) 6.5.1 EURESCOM Technical Information Network scenario and wavelength conversion alternatives An architectural model for the core of the Optical Transport Network that is having general consent is based on Optical Cross-Connects, and a meshed network topology. Moreover, legacy problems related to already deployed WDM/DWDM line terminals would probably force Network Operators to use Optical Line Terminal systems among the OXCs. This would result in savings on hi-cost Very/Ultra Long Haul WDM/DWDM interfaces, the latter being equipped only on the OLTs. The OXC could be equipped with single channel interfaces (grey or coloured) performing with 3R functionality (such as Very Short Reach interfaces) or not. This scenario is depicted in Figure 13. Referring to the simplified network scenario considered, three configurations can be identified: the OLTs connected to the OXCs are equipped with transponders that provide conversion; the OLTs have no transponders and the OXCs have to provide conversion; the OLTs and the OXCs have no transponders and conversion is provided by dedicated “ converters”. OLT OLT OXC OXC OLT OLT O L T O L T OXC O L T OXC O L T O L T OXC O L T Figure 13 Network scenario: meshed topology with OXCs and OLTs 6.5.1.1 Wavelength conversion performed by the OLTs This solution is depicted in Figure 14. The OLTs are equipped with transponders: they (de)multiplex the incoming WDM signal and each OChi associated with a specific (coloured) i (i=1…4) is transported toward the OXC over a “grey” interface. The OXC is equipped with grey interfaces only, each of them carrying an Optical Channel: if a conversion has to be done, the OXC cross-connects the OCh that has to be moved from x to y to the output port that is connected to the suitable OLT transponder. The advantages of this solution are: the OXC is equipped with grey interfaces only, thus transverse compatibility can be obtained without great efforts and it is possible to use OLTs and OXCs of different Vendors; EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 63 (93) the transparency of the Optical Cross-Connects is not an issue, any solution is feasible: pure photonic matrix surrounded by 2R or 3R interfaces, electrical matrix surrounded by 3R interfaces; OCh1_ OCh1_ OCh2_ OCh4_ OCh1_1 OCh1_1 OCh2_2 OCh3_3 OCh4_4 OCh2_2 OCh1_1 OCh1_1 OCh4_2 OCh2_3 OCh3_4 OCh4_2 OCh2_ OCh3_ OCh3_3 OCh2_3 OCh4_4 OCh3_ OCh4_ OLT OXC OCh3_4 OLT Transponder with an O/E/O conversion Single channel grey interface (eventually with an O/E/O conversion and 3R functionality Figure 14 Wavelength conversion performed by the OLTs 6.5.1.2 Wavelength conversion performed by the OXCs In this case (depicted in Figure 15) the Optical Line Terminal are not equipped with transponders and they act only as (de)multiplexers. The OXCs have now to provide the switching/routing capabilities in the network and also perform the conversions. The disadvantages of this solution are: the transparency of the Optical Cross-Connects is an issue: the OXC matrix would hardly be pure photonic. In fact, as already underlined conversion nowadays is assumed to be performed by an O/E/O conversion: considering that a signal at electrical frequencies would be available, most probably the OXC matrix would be electrical as well and consequently a 3R processing would be performed; in spite of a complication in OXCs architecture no reduction in the number of 3R interfaces is obtained; transverse compatibility could be achieved only by using expensive Inter-Domain coloured interfaces, so OLTs and OXCs would be probably forced to come from the same Vendor. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 64 (93) EURESCOM Technical Information OCh1_1 OCh1_1 OCh2_2 OCh4_2 OCh1_1 OCh1_1 OCh2_2 OCh3_3 OCh4_4 OCh2_2 OCh1_1 OCh4_2 OCh2_3 OCh3_3 OCh3_3 OCh1_1 OCh4_2 OCh2_3 OCh3_4 OCh2_3 OCh4_4 OCh3_4 OCh4_4 OLT OXC OCh3_4 OLT Single channel colored interface Single channel colored interface with O/E-E/O conversion and 3R functionality Figure 15 Wavelength conversion performed by the OXCs 6.5.1.3 Wavelength conversion performed by dedicated equipment In this case (depicted in Figure 16) neither the Optical Line Terminals nor the OXCs provide conversion: the OXC matrix is pure photonic and no O/E-O/E conversions are needed either on the cross connects or on the line terminals. conversion is performed by a dedicated equipment either by an O/E/O conversion or by analogue processing of the incoming signal. The connections among the OXC’s ports and converter are permanent. If an Optical Channel has to be cross-connected without any need for conversion, then the OXC simply operate the switching on an output port working at the same and connected with the desired OLT. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 65 (93) 4/2 3/2 1/2 4/1 3/1 2/1 OCh1_1 OCh1_1 OCh2_2 OCh4_2 OCh1_1 OCh1_1 OCh2_2 OCh3_3 OCh4_4 OCh2_2 OCh1_1 OCh4_2 OCh2_3 OCh3_4 OCh4_2 OCh3_3 OCh2_3 OCh3_3 OCh4_4 OCh1_1 OCh4_4 OCh2_3 OCh3_4 OCh3_4 OLT OXC OLT 3/4 2/4 1/4 Colored interface 4/3 2/3 1/3 Figure 16 Wavelength conversion performed by dedicated equipment On the other hand, if conversion is required then the OXC switches the signal to an output port that is (permanently) connected to the suitable converter. The output of the converter is (permanently) connected to one of the OXC ports working at the new . Finally, the signal is switched on an output port working at the same and connected with the desired OLT. If conversion is needed, the optical cross connect management system/control plane has to take care of two cross-connections: the first one to send the incoming signal to the output port connected to the right converter, the second one to switch the signal operating at the new to the output port (matching the new ) and corresponding to the desired routing. The disadvantages of this solution are: 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 66 (93) EURESCOM Technical Information the structure of the node gets extremely complicated and great care has to be taken in order to realise the proper permanent connection among OXCs and converters; transverse compatibility could be achieved only by using expensive Inter-Domain coloured interfaces, so OLTs and OXCs would be probably forced to come from the same Vendor; OXCs and converters would probably have to be provided by the same Vendor independently on how conversion is performed (either by O/E/O conversion or by analogue processing); the OXC matrix is actually passed through twice in case of conversion, the insertion loss of the OXC matrix could become critical: 3R interfaces could be needed; switching an optical channel actually implies two cross-connections, thus doubling missconnection probabilities. One possible advantage of this solution is that, being the conversion performed by separated and dedicated equipment, it could be possible to perform the conversion also in the analogue domain. In this case, parameters such as the insertion loss of the converters have to be regarded and would contribute to the already critical insertion loss of the OXC matrix (as already noted): again, 3R processing could be necessary, thus vanishing any advantage in optical conversion. 6.5.1.4 Conclusion The first solution presented best fits all network needs. In fact, it provides a complete decoupling of switching/routing elements and conversion/multiplexing elements, thus allowing deployment of any kind of OXC in the network independently on the OLT characteristics. It is also simple in terms of equipment configuration and cost effective because the OXC would be equipped only with grey interfaces. This would also allow transverse compatibility and savings in equipment cost for a Network Operator because of the multi-vendor environment. 6.6 Optical transparency in the OTN layer: advantages and drawbacks 6.6.1 Advantages One obvious advantage of transparent network elements is that, in most cases, they cost less than opaque network elements. To perform the OEO conversions on signals passing through a network element, an opaque unit requires at least a receiver, a transmitter, and associated electronics for each such pass-through signal which require extra investment. Another big advantage of transparent network elements is that they function independently of the data rate or format of the signals passing through them. This is a very attractive trait when introducing a new data rate or format. When this happens, one needs only to upgrade the network elements that generate or terminate the new signals. The core of the network needs not to be modified. And there is no limit to the bit-rate if the optical fiber can accommodate it. 6.6.2 Drawbacks Unfortunately there are some serious drawbacks to transparency. Most of them concern the control of the signal passing through the elements. Physical impairments can happen and signals impairments, such as reductions in the optical signal to noise ratio, or chromatic and polarisationmode dispersion, fibre non-linearities can accumulate from span to span. This will add an obligation to insure that, for a network to maintain a required signal quality, it is necessary to engineer it completely before rolling it out. By contrast, an opaque network can be engineered smoothly span by span as it grows. But even if building a complete optical transparent network could be done without incurring excessive costs, transparency makes future network upgrades more complicated and more expensive. In any rate, this is the case in the present context. But that doesn’t EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 67 (93) preclude, in anyway, any breakthrough which would tip the balance in favour of transparent network in the future and which could render the regeneration unnecessary or optically feasible. But for the present time, even if vendors could deliver scalable all-optical networks, practical issues may discourage carriers from rolling them out in real life. You can get transparency only at the expense of manageability and performance. Moreover, multi-vendor inter-operability will be limited if not wholly impractical. In addition, in transparent network it is difficult to monitor the quality and identity of the individual signals. As a result, network management and fault location are more complicated and all the more so expensive than for opaque networks. Restoration will be limited to 1+1 case. On the other hand, opaque networks can implement the best technology available in each system that is added with no barrier to interconnecting systems with different technologies. This shows that opaque networks scale more gracefully. Opacity has one drawback. It requires numerous transponders, which need to be changed if the bit-rate has to be upgraded. Transponders do not come for free which make them less ideal. 6.7 Conclusion ASON should provide customers with several new services thanks to its automatic functions implemented in its control plane. The basic functionality of ASON is defined for providing these services and is described in Chapter 2. Since ASON can be either opaque or transparent or a mixture of both, the services to be provided by ASON are independent of ASON's optical transparency issues. From the point of view of the functionality, both opaque and transparent ASON should perform the same set of basic functions that are determined by the services to be provided. We have found that a fully transparent ASON requires additional functions, extended database, or appropriate network solutions – some of which are not available, yet – compared to opaque one. In summary, an optically transparent ASON would be very complicated to manage and thus does not seem feasible. Regarding a fully transparent OTN layer without wavelength conversion, it is also pointed out that impairments of optical signals due to a number of physical effects will severely limit the length of a transparent span and could limit the size of the network. There are some other, related problems, such as resource allocation and wavelength blocking. However, these could be mitigated by introducing wavelength conversion or optical transparent islands. Our conclusion is, that although optical transparency could provide some important advantages at the OTN level, but these can not compensate the drawbacks at the control plane level, and thus a fully transparent ASON is not attractive (might not even be feasible) for the time being. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 68 (93) EURESCOM Technical Information 7 Compatibility of MPS with ASON 7.1 Introduction Optical network elements and optical networks are newcomers in the telecommunications industry. MPS, which is a part of Generalised Multi-Protocol Label Switching (GMPLS), has been proposed for the control plane of these. Traditional circuit switched transport networks have virtually no automatic network control, and as a result very slow service provisioning. Automatically switched transport networks are required to make capacity easily available to clients. Automatic switching of available capacity is of paramount importance to quickly and efficiently provision bandwidth wherever and whenever needed to satisfy the growing demands for network bandwidth. A lot of attention has been focused recently on new control plane architectures based on the adaptation of data networking protocols and their extensions to the optical domain. The potential of these new architectures for optical networking are huge as demands for bandwidth continue to increase rapidly thanks to the explosive growth of the internet and the proliferation of virtual private networks (VPNs). A control plane architecture should provide service providers with enhanced network control capabilities and relieve operators from unnecessary and costly manual operations. At the same time, it should facilitate network interoperation between different networks with different transport technologies such as circuit switched networks and data networks. Also, equipment from different vendors employing different technologies must inter-work at the control plane in order to preserve the investment made by operators. As mentioned in [13], GMPLS, which is an extension of MPLS to the time and optical domain, combines recent advances in MPLS traffic engineering control plane constructs with optical network element technology, with the aim to: Provide a framework for real time provisioning of optical channels in ASON Foster the expedited development and deployment of a new class of versatile optical cross connects (OXCs) Allow the use of uniform semantics for network management and operation control in hybrid networks that consist of OXCs and label switching routers (LSRs) Chapter 6 aims at clarifying whether MPS (GMPLS) is suitable for the control plane of ASON. This section starts with a short introduction to MPLS, and proceeds to identify the main considerations regarding the extension of MPLS to the optical domain. An introduction to GMPLS is then presented that is followed by a summary of the main features that are required for the ASON control plane. Subsequent sections cover routing, traffic engineering, link bundling, protection, restoration, signalling, as well as handling of control plane failures in GMPLS. Finally, the activity is concluded with a discussion of the compatibility /suitability of MPS for the control plane of ASON. 7.2 MPLS MultiProtocol Label Switching (MPLS) is the latest technique that is able to provide virtual path capabilities across Label Switched Routers (LSR) based on labels [12]. This technique allows carrying efficiently different services across an IP network by performing traffic engineering to avoid congestion and use all the available network resources. MPLS can be seen as a Layer 2.5, that is, a layer under IP and over any Layer 2 such as ATM, SDH or Frame Relay. MPLS adds a header (called MPLS header) between the IP and the Layer 2 header (see Figure 17). The MPLS header has some fields such as a label (or stack of labels) and the TTL (other fields are optional): EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 69 (93) The label is a short, fixed length, locally significant identifier, which is carried by the packet and is used to identify a Forwarding Equivalence Class (FEC). The TTL is the Time-To-Live of the packet, which is used to keep the right value after traversing an MPLS network. Figure 17 MPLS Header The MPLS nodes called Label Switched Routers (LSR) can be either IP routers or ATM switches that have been enhanced to perform MPLS signalling and routing. 7.2.1 Label operations The operations that can be perform with the labels of the packets are four (Figure 18): Label add/drop: it is performed at the edge nodes: the label adding at the LSP origination and the label dropping at the LSP termination. At the LSP origination, the label is assigned depending on the FEC where the packet belongs. In MPLS networks, only at the edge node the FEC is checked. After that, the switching is based only in the label. Label swapping: This is the basic forwarding operation consisting on looking up the incoming label of the packet in order to determine the outgoing label and port that the packet will have. Figure 18 Label swapping example: data that reaches the core LSR with Input Port (IP) 0 and label (IL) 3 will be forwarded to Output Port (OP) 2 with Label (OL) 5. Label merging: This operation is done when two or more separate labels are combined into a single one in an aggregated manner so that the number of used labels is reduced and the packets going to the same egress node and belonging to the same FEC use the same label although they come from different ingress nodes. Label stacking allows path aggregation by pushing a new common label onto the label stack of the MPLS packet. This operation is useful to have MPLS clouds within the MPLS network. 7.2.2 Label allocation The labels can be allocated downstream on demand or unsolicited-downstream. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 70 (93) EURESCOM Technical Information Downstream-on-demand allocation is when there is an upstream LSR that has a data flow and requests for a label to the next LSR. Depending on the upstream node and the FEC of the flow, the label will be allocated only for the requesting LSR. Unsolicited-downstream allocation is when the downstream node allocates labels and distributes the label to all its adjacent neighbours. Its neighbours have to decide whether they accept this label and include it in their switching table. It should be mentioned that the switching tables and the label allocation will depend on whether the LSR is able to distinguish between interfaces. That is, in most LSRs, labels are unique on a per interface basis but in some cases, labels are unique on a per LSR basis and therefore different labels must be used for packets coming from different interfaces. 7.2.3 Label distribution and signalling The label distribution protocol is the set of procedures by which one LSR informs the others about the label/FEC it has made. There is no a single label distribution protocol. There are extensions of existing protocols such as MPLS-RSVP or MPLS-BGP and there are new protocols such as MPLSLDP. To date, two protocols are the most preferred: RSVP and CR-LDP RSVP is a layer 3 signalling protocol independent of the routing protocol (it works with shortest path, QoS routing and constrained routing). Path control message is sent by the source and the reservation is receiver initiated and unidirectional informing of the requested QoS parameters (BW, max. delay…). CR-LDP is a set of procedures and messages by which LSRs establish Label Switched Paths (LSPs) satisfying an arbitrary set of constraints. 7.2.4 Aggregation Path aggregation: MPLS is able to bind a single label to a union of FECs, which is a single FEC within some domain. In this way, packets from different ingress nodes use the same label and follow the same path. The advantage of path aggregation is that it reduces the number of labels needed in the MPLS network. 7.2.5 Routing IP networks are best-effort networks that do not know about network balancing and do not avoid congestion. MPLS wants to avoid these limitations by being able to choose other routes than the shortest ones. For this purpose, MPLS supports both explicit and constrained based routing. Explicit routing allows forcing the packet to follow a certain path. In IP, explicit routing is possible but it adds a lot of overhead because the addresses of all the routers in the path have to be specified within the header of each packet. MPLS facilitates explicit routing so that each route is related to one label and the packets that have to follow this path will be assigned to it. Constrained based routing: constraint-based routing algorithms are able to find a path that optimises a certain scalar metric but at the same time it does not violate a set of constraints (e.g., when aiming to find a path with a minimum available bandwidth, the constraint could be that all the links along the path should have at least the requested available bandwidth). No routing protocol has been specified in MPLS but extensions of the protocols OSPF and IS-IS are mostly being used. 7.2.6 Traffic Engineering Traffic Engineering is concerned with the performance optimisation of the network by controlling the network resources. IP does not offer traffic engineering because it is a best effort based network where the only performance improvement can be done by TCP. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 71 (93) On the contrary, MPLS is able to offer traffic engineering by using constraint-based routing. An MPLS network given its topology and the traffic matrix, its able to find which is the route offering the best overall network performance. There are several tools mainly based on statistics. An example of a tool offering MPLS TE is NetCalc by Nortel Networks. There are two types of constraints considered in traffic engineering: Restrictive constrains as the minimum bandwidth required by the connection, Additive constrains are the parameters, which are added along the path, as for example the delay or the jitter. The parameters taken into account for traffic engineering are mostly bandwidth, set-up and holding priorities (used in Admission Control), resource class affinity and path options (for routing selection). 7.3 Introduction to GMPLS IETF is currently working on an adaptation of MPLS for the optical layer, which is called Multi Protocol Lambda Switching, and which was of late translated to Generalised MPLS (GMPLS), to achieve automatic connection control also called MPlamdaS or MPS. It is the extension of MPLS to optical networks. 7.3.1 Main issues involved in the extensions of MPLS to encompass the optical domain (MPS) One of the inherent reasons why WDM can provide big networking advantages is the fact that it is relatively easy to build wavelength dependent network elements. Wavelength dependence is a standard property with most optical elements: an aspect that complicates transmission systems but also simplifies multiplexing and de-multiplexing, add-drop functionality, detection and routing. Moreover, the wavelength information is in a different domain (optical) than the signal information itself (electrical). Optical add-drop multiplexing, de-multiplexing and cross-connection can find place based on the wavelength of the signal combined with the port it arrives at without the need for reading the content of the signal. This aspect makes wavelength the perfect label because it adds no overhead to the signal. A cross-connected WDM optical network can provide end-to-end optical (OCh, lightpath) connections between node pairs, and an OXC can be configured to direct a signal with a given wavelength from a certain input to a certain output. The analogy between MPLS networks and optical wavelength routed networks is rather clear and is shown in Table 6. Equivalence between MPLS and Optical Wavelength Routed Network MPLS Optical Wavelength Routed Network (MPS) Label Switched Path (LSP) OCh or Lightpath Label Switching Router (LSR) OXC Selection of Labels Selection of ’s and OXC ports Table 6 At the same time, since MPLS was not originally created with optical technology in mind, there are certain aspects of it that either cannot be realised in optical networks or are difficult to realise in optical networks – and therefore ought to be best avoided. Adaptations or changes of thought are therefore needed in addition to the extension in definitions. Some of the aspects of optical networks that do not quite correspond to MPLS are listed below: Label merging is not straight forward to realise optically; 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 72 (93) EURESCOM Technical Information Label pushing and popping appears difficult/ expensive to do optically; Label stacking is difficult to realise optically; Label swapping is possible but expensive to realise optically; Optical channels have traditionally been bi-directional. On the other hand, optical technology provides also new possibilities as a number of label-related functions are easier to perform optically than electrically. This simplifies the overall architecture since some of the functions required in MPLS label selection/ distribution become superfluous in an optical network. Some of these aspects are listed below: End-to-end wavelength paths can be realised in an optical network – thus eliminating to an extent the need for label swapping; The label and the signal payload are in two different domains (optical and electrical respectively) such that they can be easily separated; The OXC port is an aspect of the optical label that can be pushed, popped, swapped and stacked [13]. Figure 19 Label hierarchy in GMPLS 7.3.2 Generalised Label The extension of MPLS to GMPLS aims at achieving a more efficient labelling and forwarding mechanism by using a generalised label that is applicable to different technologies and networks. A single set of control plane protocols with enough flexibility can support different types of networks. For IP routers the labels designate principally input and output ports. For optical networks they designate input and output ports as well as wavelength or band of wavelengths for each OXC. For the SDH ADMs and digital cross-connects, they designate input and output time slots. Here labels can be time-slots, wavelengths, wavebands, and fibres. Labels cannot be stacked but form a hierarchy, which is shown in Figure 19. On the contrary, when use of hierarchical labels is EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 73 (93) needed, LSPs have to be established separately [14]. The information carried within the generalised label depends on the type of link where the label is used. 7.3.3 Label requests Several drafts have been presented to give the information that should be exchange when requesting labels. This information is shown in Table 7. LSP Encoding Type Link Protection Flags G-PID 8 bits 8 bits 8 bits Packet PDH Ethernet SDH SONET Lambda Etc. Unprotected Shared 1:1 1+1 etc. DS1 SF DS1 ESF DS3 M23 ATM mapping VT Etc. Table 7 Generalised Label request information [14] G-PID: Generalised Payload Identifier; DS1: Digital Signal level 1; SF: Super Frame; ESF: Extended SF; VT: Virtual Tributary. 7.3.4 Generalised LSRs In this architecture, there are several kinds of LSR depending on the forwarding they perform and their interfaces: Interfaces able to recognise the packet boundaries and their forwarding is based on the packet header (as classical MPLS LSRs); Interfaces forwarding based on the time slot (so-called TDM interfaces); Interfaces forwarding based on the incoming wavelength (so-called Lambda Switch capable interfaces); Interfaces forwarding based on the incoming physical port (so-called Fibre Switch capable interfaces). 7.3.5 Architecture models There are several models, which can be applied to GMPLS: The overlay model consists on having the optical control plane and the client control plane separated and hence, the signalling and routing protocols at the router network are independent from the signalling and routing protocols of the optical network; The augmented model consists on having still separated control planes at the optical and client layers but there is some exchange of routing information between them; The peer model consists on sharing the same control plane by both, the optical and the routing plane. A particular case of GMPLS could be the Automatic Switched Optical Network (ASON) when using the MPLS signalling and routing protocols. The more integration there is at the control routing plane, the more different signaling is needed. 7.3.6 Bi-directional LSPs A new feature of GMPLS networks is the possibility of establishing bi-directional LSPs. A bidirectional LSP consists on two unidirectional LSPs with the same traffic engineering, LSRs, protection and restoration, and resource requirements in each direction. In this case, two labels 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 74 (93) EURESCOM Technical Information must be allocated. In order to reduce probability of contention, there is a mechanism to suggest labels to the downstream direction. Also there is a mechanism to resolve the contention problem by giving priority to the node with higher ID. 7.4 On the suitability of MPS for the ASON control plane The compatibility between ASON and MPS was still an issue at the forming phase of P1012. This issue has meanwhile been clarified within the international community, driven especially by work within IETF, as well as OIF and to an extent ODSI. GMPLS is not simply compatible with ASON, it is in fact probably the strongest candidate for the control plane of ASON as it aims to combine the best of two worlds: namely the “intelligence” of MPLS with the added value of the optical label and the transparency of the optical network. Extensions and adaptation to GMPLS are expected to be an on-going process for a while as optical technology is still undergoing an explosive growth. One of the main reasons why MPS is probably the strongest candidate for the ASON control plane stems from migration considerations. There are indeed a range of architectural options regarding the control plane of the optical NEs and network. ASON is the option – overlay model – where intelligence is owned by the optical network. The opposite extreme, and the favourite of the IP world, is that there is no optical network as such, just unintelligent optical network elements that are extensions of the IP routers. This is the peer-to-peer model. The third alternative is actually a whole range of alternatives that range from the ASON model to the peer-to-peer model that are called in unison the augmented model. Here some minimal routing information is exchanged between the optical network and its client(s). The three models are clearly alternative architectures that are supported by different vendors depending on whether the vendor has been traditionally a main actor in IP or in circuit switched networks. However, it has now become rather clear that the three models are not directly competing for the time being in the sense that the equipment and protocol adaptations needed for the implementation of the peer-to-peer model limit its applicability in the short term. On the other hand, it is expected that some “heavily peered” augmented model will dominate after a while when IP becomes the main network technology. Rather than competing, the three models are now seen as evolution steps down one single network evolution curve – as schematically shown in Figure 20. ASON PEER-to-PEER Figure 20 Evolution scenario of vertical network architecture The peer-to-peer model and the augmented model will opt for MPS control planes. It is for this reason most efficient that MPS is also used in the control plane of ASON as this will minimise development work and will allow a seamless evolution of the network. 7.5 Summary MPS is an extension of MPLS to optical networks and is part of GMPLS. The use of a generalised label makes GMPLS - currently under development - multi-client. GMPLS has many of the characteristics that are required for the ASON control plane. It can provide efficient routing, signalling, and traffic engineering mechanisms. It can thus provide the optical transport network with the intelligence that is required to realise ASON. We contend that GMPLS is the strongest candidate for realising the control plane of ASON, primarily because of its relative maturity combined with its compatibility with IP traffic and networks. This strengthens the position of MPS for implementation in the ASON control plane. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information 8 page 75 (93) Conclusions ASON is based on the overlay model and should provide its different clients with the OCh services defined in [1]. Several of them are new OCh services that are provided by its automated connection handling capabilities. These services are applied in backbone networks. Functionality of ASON is defined for carrying them and can be defined as the basic functionality. This category of functionality is invariant of many other considerations, like optical transparency issues, since it is only related to OCh services provided by ASON. Other functionality, if it exists, essentially depends on the technological solutions used in implementing ASON. Therefore discussions on different aspects of ASON should be done keeping these boundary conditions in mind. The first, we provided a description of the functionality from two perspectives: core-edge and client-server. The core functionality that is mainly associated with switching of optical channels includes topology discovery, optical routing, signalling, policing and CAC. As for those of the edge that deal with various controls of incoming and outgoing traffic streams include QoS handling, connection handling, CAC, policing and UNI interface functionality. These functions cooperate to perform automated end-to-end OCh processes such as provisioning, protection or restoration. Functionality of the UNI interface allows some Clients to request a connection directly. The description of functionality required for handling QoS revealed a lack of definition for OCh services. Nevertheless, QoS handling issues should be addressed at the edge nodes. From the client-server aspect, the server layer functionality can be grouped into control plane functionality and OTN layer functionality. The former is very similar to the functionality described from the core-edge perspective and the latter refers to the native functionality of the OTN layer such as optical cross-connection. We have attempted to create a short summary of the Client functionality in order to determine the Client-ASON relationship at the functionality level. We have included Frame Relay, ATM, SDH and IP into our investigation. The results show a strong similarity in the functionality between ATM with PNNI protocol and MPLS-TE based IP and the completeness of their functionality for handling OCh services. A study on the interworking of ASON with its Client layers suggests that ASON could perturb its Client layers' activities during its reconfiguration phase, for instance, after a failure. However, no particular implementation issues are expected at the functionality level. Chapter 3 discusses the specific management requirements for the ASON control plane. These include the management of the topology (i.e. node, link and path), end-to-end paths, control channel, protection/restoration and UNI interfaces. Also a comparison is provided of the distributed management approach of ASON and that of conventional centralised management, such as TMN. The advantages and disadvantages of each management approaches are summarised together with their architecture schemes. Please note that the general management requirements currently applied to any telecommunication network also apply to ASON. Chapter 4 discusses grooming. Two grooming schemes (ASON-SDH and ASON-OTN) are considered as grooming at the ASON edge. ASON-SDH refers to the case where both core and edge nodes have grooming capabilities thanks to SDH multiplexing functions. Such ASON could directly deal with sub-OCh rate streams. Whereas in the second one, ASON handles only OCh rate traffic streams and the incoming client traffic streams are groomed at the edge. We propose a grooming scenario that is based on the ASON-OTN scheme. In this scenario the G.709 digital wrapper is used. SDH multiplexers and IP routers with POS or (10) GbEth interfaces provide additional possibility for grooming. Chapter 5 is devoted to the different aspects of optical transparency. From the point of view of the functionality, both opaque and transparent ASONs should perform the same set of basic functions that are determined by the services to be provided. We have found that a fully transparent ASON requires additional functions, extended database, or appropriate network solutions – some of which 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 76 (93) EURESCOM Technical Information are not available, yet – compared to opaque one. In summary, an optically transparent ASON would be very complicated to manage, and thus does not seem feasible. Regarding a fully transparent OTN layer without wavelength conversion, it is also pointed out that impairments of optical signals due to a number of physical effects will severely limit the length of a transparent span and could limit the size of the network. There are some other, related problems, such as resource allocation and wavelength blocking. However, these could be mitigated by introducing wavelength conversion, optical transparent islands or hybrid nodes. Our conclusion is, that although optical transparency could provide some important advantages at the OTN level, but these can not compensate the drawbacks at the control plane level, and thus a fully transparent ASON is not attractive (might not even be feasible) for the time being. Chapter 6 focuses on the compatibility of MPS with the ASON control plane. The question was triggered by an IP-centric approach to ASON in which IP protocols are used to control and manage optical networks. MPS is an extension of MPLS to optical networks and is part of GMPLS. The use of a generalised label makes GMPLS - currently under development - multi-client. GMPLS has many of the characteristics that are required for the ASON control plane. It can provide efficient routing, signalling, and traffic engineering mechanisms. It can thus provide the optical transport network with the intelligence that is required to realise ASON. We contend that GMPLS is the strongest candidate for realising the control plane of ASON, primarily because of its relative maturity combined with its compatibility with IP traffic and networks. This strengthens the position of MPS for implementing the ASON control plane. In conclusion, we have defined: the ASON functionality required for carrying OCh services together with implementation issues; specific management requirements for the ASON control plane; a grooming scenario at the edge; In particular we have highlighted: the impact of optical transparency; the compatibility of MPS with the ASON control plane. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 77 (93) References [1] EURESCOM Technical Information "Network requirements and scenarios" of project P1012 FASHION (EDIN: 0199-1012), 2001. [2] ITU Q19/13 First draft of G.ason, "Architecture for the automatic switched optical network", September 2000. [3] oif2000.155.1, "Carrier optical services framework and associated requirements for UNI", OIF, September 2000. [4] ITU COM 15-32-E, "ASON characteristics", December 2000. [5] D. Awduche et. Al., "Requirements for traffic engineering over MPLS", RFC-2702, September 1999. [6] EURESCOM Project P918 "Integration of IP over Optical Networks: Networking and Management", Deliverable 3: "Optical Transport Network Management", 2000. [7] "Architecture of optical transport networks", ITU-T Recommendation G.872 (02/99). [8] "TMN management functions", ITU-T Recommendation M.3400 (04/97). [9] ECOC’99 proceedings, “High-speed transmission over standard fibre” [10] Final report of Action COST 239 [11] "Transmission capacity of optical path overhead transfer scheme using pilot tone for optical path network", J. Lightwave Technol., vol.15, no.12, dec. 1997. [12] MPLS: Technology and Applications, by B. Davie and Y. Rekhter, Morgan Kaufmann Publishers, Academic Press 2000. [13] Multi-Protocol Lambda Switching: Combining MPLS traffic engineering control with optical cross connects: draft-awduche-mpls-te-optical-02.txt [14] Generalised MPLS Signalling Functional Description: draft-ietf-mpls-generalisedsignaling-01.txt 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 78 (93) EURESCOM Technical Information Annex A QoS A.1 IP Quality of Service Traditionally, IP has supported best effort service only, where all packets have equal access to the network resources. The classical IP network framework does not handle the critical issue of Quality of Service (QoS) at the IP layer, with the exception of the Type of Service (ToS) field in the IPv4 packet header format, allowing QoS to be controlled at a different level. This approach has already been ruled as inadequate and has therefore changed for the IPv6 specification. The new IP network, based on the IPv6 will intrinsically support QoS functions at the IP layer, making for a much more robust and performance-centric approach. Lately IETF has come with several solutions to enable QoS in the Internet. Among these, the IntServ/RSVP and the DiffServ/QoS-agents are promising models. An alternative approach to QoS is via MPLS and is presented in the next section together with the other features of MPLS. The discussion concerning IP network QoS functionality should not be limited to the specifics of the IPv6 implementation, but should also take into account the IPv4 approach, consisting of such protocols and classification methods as RSVP signalling, DiffServ and IntServ. These are relevant when considering the case of IPv4/IPv6 hybrid networks and the necessity for protocol independent QoS handling. Furthermore since (as mentioned also in the introduction) we expect that for some time IPv4 will dominate at the edge while IPv6 at the core of the network it is important to consider hybrid client layers and what effect this will have if we are to offer end-to-end QoS (similar to the one promised by ATM). A.2 DiffServ: The Differentiated Services Model Differential service mechanisms allow providers to allocate different levels of service to different users of the Internet. Each corporate or ISP network is a DiffServ domain. Within a domain, the traffic and packets are handled in a common manner. The IETF’s DiffServ CodePoint (DSCP) in the packet header defines a per-hop behaviour. Traffic entering a network is classified and assigned to different behaviour aggregates. Each behaviour aggregate is identified by a single DSCP, which is contained in the packet header. Within the network, packets are forwarded according to the perhop behaviour associated with the DSCP. There are DiffServ boundary nodes that classify the traffic and mark it appropriately. Between domains, Service Level Agreements (SLAs) and Traffic Condition Agreements (TCAs) are used. This means that Diffserv does not provide any mechanism for reserving resources in the network, and in many networks this will probably mean that DiffServ will only work as a mean for providing CoS. However, how DiffServ will be used is still under discussion. DiffServ offers QoS to traffic aggregates by using some functional elements implemented at the nodes. These include: A set of per-hop forwarding behaviours, which define the classes of QoS offered. Classification of each incoming packet by marking the DS-field of the packet header (6 bits of TOS and TC fields of IPv4 and IPv6) to associate it with a particular per-hop behaviour aggregate. Traffic conditioning, which includes metering, dropping and policing. Packet classification and traffic conditioning are performed at the edge routers only. In the core network, DiffServ relies on classification of the fixed length DS-field only. This makes DiffServ a much more scaleable model. Two types of packet classifiers are defined in DiffServ: classifiers that are based on the DS-field only and, Multi Field (MF) Classifiers which select packets based on the value of the combination of source address, destination address, DS-field, protocol ID, source port and destination port number. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information A.3 page 79 (93) IntServ: The Integrated Services Model The Resource Reservation Protocol (RSVP) and architecture for realising end-to-end guaranteed QoS are the results of work in the IntServ group. RSVP is a signalling protocol to establish and maintain network resource reservation. RSVP has thus a set-up phase where resources are reserved in every intermediate router (such as bandwidth or CPU power). IntServ defines the services and, together, the two enable users to reserve for unicast and multicast flows between QoS aware applications. If every hop agrees to reserve sufficient resources, the user will have a dedicated reserved stream of guaranteed capacity. Once the session ends, the reserved resources are released. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 80 (93) EURESCOM Technical Information Annex B MPLS B.1 Multi Protocol Label Switching (MPLS) integration MPLS (Multi-Protocol Label Switching) is by definition a protocol independent of the IP framework, but it should be discussed and analysed from the IP network perspective, as it is expected that this interoperation between the IP and the MPLS will become the norm in future internetworking. Although MPLS is originally proposed as a solution to improve on the limitation of IP routing (discussed above), the move to MPLS is dictated by the new needs that arise and cannot be directly addressed by the current IP network framework. The MPLS provides solutions to such issues as IP traffic engineering, easy to implement QoS, networking security, VPN deployment, etc., while offering seamless and tight integration of the IP Layer and the underlying Layer2 network technologies. Furthermore, the basic concept of the MPLS seems to be appealing in the case of optical networks, for which the MPλS has been proposed. Additionally, it appears that there might be significant duplication of functionality when IP/MPLS is to be used as the client to an ASON network. B.2 MPLS principle: IP packet classification into packet flows A flow is a packet sequence for which the same routing decision is applied for every packet. An example of a flow is a packet sequence with the same source and destination IP addresses. Furthermore, classification of IP packets into flows maybe based on the packet source and destination address/port (TCP or UDP) combination. In this way packets that belong to different applications may be handled in different ways in order to satisfy particular QoS requirements. The forwarding mechanism for each flow is usually determined when handling the first packets that belong to the specific flow. An example of flow classification is Tag Switching or MPLS, which is discussed in the next section. B.3 MPLS basics A number of different approaches have been suggested to “push” as much as possible of the traffic flow down to Layer 2, thus performing “switching” instead of “routing” in IP-networks. MPLS is an effort from the IETF to create a standardised solution for this. The “label” is a number (20bitlong), assigned at an IP router at the edge of a label switched or MPLS domain, which identifies a path across the network, so that the packets may be routed more quickly without having to lookup the destination address in the IP packet. This label can either be appended to the IP packet, or can be stored in the encapsulation frame when a suitable field exists. MPLS is not restricted to any link layer technology and may use forwarding provided by ATM or Frame Relay equipment. In MPLS IP packets are classified in so-called Forwarding Equivalence Classes (FECs) at the ingress of the MPLS domain. An FEC is a group of IP packets that are forwarded over the same path and treated in the same manner. The assignment may be based on e.g. host address, or a “longest match” of the destination address prefix of IP packets (thus destination). The FEC to which an IP packet is assigned is encoded with a short, fixed length label. The header is composed of a 20bit-Label, an 3bit-exp, an 8bit-TTL and a 1bit-s fields. At the nodes of an MPLS network the labelled packets are forwarded based on the label swapping paradigm. This means that the label associated with the IP packet is examined at each Label Switching Router (LSR) and is used as an index into the Label Information Base (LIB). The label is mapped on a next hop label forwarding entry in this table that determines where to forward the packet. The old label is replaced with a new label and the packet is forwarded to its next hop. Thus, while an IP packet is in the MPLS domain the packet’s network header is no subject of further analysis at subsequent MPLS hops. In order to establish and maintain a path according to the information collected by the routing protocols, the LSRs along this route must assign and distribute labels among neighbouring nodes. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 81 (93) Herewith a Label Switched Path (LSPs) is created between ingress and egress points of the MPLS domain. An LSP is created by the concatenation of one or more label switched routers, allowing a packet to be forwarded by swapping labels. The distribution of labels is required to allow a LSR to inform another LSR of the label/FEC binding it has made. With this the LIB in the LSRs used in the label swapping process is kept up to date. A Label Distribution Protocol (LDP) accomplishes the distribution of label/FEC bindings among participating LSRs, in order to establish an LSP. MPLS provides a number of benefits to IP network providers: Efficient forwarding: due to the use of labels, core routers/LSR no longer need to make a route lookup in very large routing tables, instead a much smaller LIB can be used to make the forwarding decision; Service differentiation: certain paths, or FECs, may be assigned different CoS. Use of the label in combination with CoS parameters allow easy identification of such traffic streams; MPLS Virtual Private Networks: VPNs can be created in a relatively straightforward manner. Again by using (different) labels, private traffic can be isolated in a public network; Traffic engineering: because MPLS paths are topology based, and labels are used to identify them, a path easily can be re-routed. Also here the label is used to accomplish this. It can be implemented on devices based on ATM switches, thus achieving line-speed packet forwarding. Besides the numerous advantages, there are some disadvantages as well: Granularity: MPLS only considers traffic aggregates. As a result per micro-flow QoS, or any other per micro-flow operation may be hard to realise. Note that it is possible to assign a label to each flow, but that it would not scale due to the large number of flows in the core of the network; Topology oriented: because MPLS is topology oriented, a label needs to be assigned to each route. Some judge this as a weakness, since a route may not be in use, and therefore the label would be wasted; 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 82 (93) EURESCOM Technical Information Annex C ATM network C.1 ATM basics C.1.1 ATM switching operation The basic operation of an ATM switch is as follows: 1. The cell is received across a link on a known VCI or VPI value. 2. The switch looks up the connection value in a local translation table to determine the outgoing ports of the connection and the new VPI/VCI value of the connection on that link. 3. The switch retransmits the cell on that outgoing link with the appropriate connection identifiers. All VCIs and VPIs have only local significance across a particular link. Therefore these values are remapped, as necessary, at each switch. C.1.2 ATM reference model ATM layers: ATM physical layer; This layer manages the medium-dependent transmission. Functions of this layers are: conversion of bits into cells; control of transmission and receipt of bits on the physical medium; tracking of ATM cell boundaries; packaging of cells into the appropriate type of frame for the physical medium. ATM layer; This layer is combined with the AAL and is responsible for establishing connections and passing cells through the ATM network using information in the header of each ATM cell. The ATM layer is responsible for mapping network addresses to ATM addresses. ATM Adaptation Layer (AAL). This layer is responsible for isolating higher-layer protocols from the details of the ATM processes. Planes spanning all layers: Control plane; This plane is responsible for generating and managing signaling requests. User plane; This plane is responsible for managing the transfer of data. Management plane. Layer management: manages layer specific functions, such as the detection of failures and protocol problems. Plane management: manages and coordinates functions related to the complete system. C.2 SVC connection procedure SVC is a connection that is set up through a signalling protocol with the following steps: Connection request An ATM device (source end-system) sends through a UNI interface a "Setup message" to its directly connected ATM switch (ingress switch). This request contains the ATM address of the ATM endpoint and traffic and QoS parameters required for the connection. The ingress EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 83 (93) switch sends a "Call Proceeding message" back to the source and invokes an ATM routing protocol. The signalling protocol between the ATM device and its switch is ATM UNI. Connection request routing When receiving a connection request, an ATM routing protocol, such as PNNI, computes a path based on information stored in the node's topology database, together with the connection's service category, traffic and the QoS parameters requested by the source endsystem. The signalling protocol (PNNI signalling) propagates the "Setup message" along the path given by the routing protocol across the network (through NNI interfaces). When receiving the Setup message the egress switch forwards it to the end-system across its UNI. Connection accepted/denied (Connect/Release message), The ATM end-system sends a "Connect message" if the connection is accepted. The "Connect message" traverses back through the network along the same path to the source end-system, which sends a "Connect Acknowledge message" back to the destination to acknowledge the connection. If the connection is not accepted the end-system sends "Release message" back. Information transfer, Connection release (Release message). 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 84 (93) EURESCOM Technical Information Annex D Generic OTN management requirements (after G.872) D.1 Introduction The description of requirements for the management of optical transport network can be based on G.872. For the most part, it still remains valid for the OCh management in ASON. That is, the difference would come from a set of automatic functions implemented in the control plane of ASON. This Annex recalls briefly OTN management requirements described in G.872. The description will be restricted within the OCh management. D.2 Generic management requirements D.2.1 Generic fault, configuration and performance management The optical transport network shall provide support for fault, configuration and performance management end-to-end as well as within and between administrative boundaries. It shall provide a means of detection and notification in the event of a misconnection. The optical transport network shall provide facilities to: Ensure interconnection of transport network entities that have compatible adapted or characteristic information; Detect fault, isolate faults and initiate recovery actions where applicable. The optical transport network shall provide facilities for single-ended maintenance. In the event of a signal within the server layer being interrupted, upstream and downstream network entities in the server layer shall be notified. The optical transport network shall be able to detect performance degradations to avoid failures and verify quality of service. D.2.2 Generic management communications The optical transport network shall support communications between: Personnel at remote sites; OSs and remote NEs; Craft terminals and local or remote NEs; Between (or among) NEs. These forms of communication may also be supported externally to the optical transport network. D.2.3 Generic client/server interaction management The optical transport network shall detect and indicate when a signal is not present at a client layer, within the OTN, also in the case where the server layer is operating normally. In order to avoid unnecessary, inefficient or conflicting survivability actions, escalation strategies (e.g. introduction of hold-off times and alarm suppression methods) are required: Within layer Between the server and client layer. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information D.2.4 page 85 (93) Optical channel layer network management requirements Requirements for management capabilities of the optical channel layer network are identified in this section. Connection supervision Connection supervision is a management to provide supervision of the integrity of network connections that are supporting the trails in any layer network. A link connection supported by a server layer network is supervised by means of continuity supervision. Continuity supervision This refers to the set of processes for monitoring the integrity of the continuity of a trail. The process involved in is "Detection of Loss of Continuity (LOC)". In general, the failure of a link connection in a server layer will be indicated to a client layer through some form of server signal fail indication. A server signal fail detected by the OCh trail termination sink (the signal may be forwarded by the OMS layer) will lead in turn to a server signal fail towards the client layer. The processing in the OCh adaptation source of the server signal fail is client specific. Connectivity supervision This refers to the set of processes for monitoring the integrity of the routing of the connection between source and sink trail terminations. "Trail Trace Identification" (TTI) is the process for connectivity supervision. TTI is necessary to ensure that the signal received by a trail termination sink originates from the intended trail termination source. TTI at the OCh layer is only needed where there is a possibility of channel rearrangement between OCh source/sink trail terminations. Maintenance indication This refers to the set of processes for indicating defects in a connection being part of a trail. The defect indications are given in downstream and upstream directions of a bidirectional trail. The following two maintenance indication processes enabling defect localisation and single-ended maintenance are involved in. : - Forward Defect Indication (FDI) - Backward Defect Indication (BDI) FDI is used to indicate downstream that a defect condition has been detected upstream. BDI signals the state of the trail at the trail termination sink back to the remote trail termination sink. Signal quality supervision This refers to the set of processes for monitoring the performance of a connection being supporting a trail. This is necessary for determining the performance of connections. Generic processes include parameter measurements, collection, filtering and processing. Signal quality supervision is needed to manage channels and multiplexed channels at the network level management. Thus performance parameter monitoring is required at the OCh and OTS layers. Adaptation management This refers to the set of processes for managing client layer network adaptation to/from the server layer network. "Payload Type Identifier (PTI)" process is necessary to ensure the client layer is assigned at connection set-up to the appropriate source and sink OCh/Client adaptations. A PTI mismatch detected at source or sink adaptations would indicate an incorrectly provisioned or altered client-OCh server layer adaptation. Protection control This refers to the information and set of processes for providing control of protection switching for a trail or sub-network connection. Protection switching is controlled on the basis of local criteria 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 86 (93) EURESCOM Technical Information generated by the trail or sub-network connection supervision and by the TMN/OS. Control from the remote network element using an automatic protection switching protocol (APS) is possible too. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information Annex E Optical transparency in the OTN E.1 Technology related issues E.1.1 Optical amplification page 87 (93) In this area, optical transparency has had a big success. In the opaque regenerator, the input multiwavelength signal is first de-multiplexed into its individual channels. For each channel one performs an OEO conversion and they are then multiplexed together to form the former multiwavelength signal. On the other hand, The transparent optical amplifier amplifies all channels, without having to de-multiplex and multiplex them again. Even if optical amplification doesn’t solve all problems, it offers reasonable robust solutions and that shows why it is widely used. But one has to engineer the total path so as to maintain adequate optical SNR at the endpoint. This requires amplifiers with flattened and regulated gain spectra. A major hurdle is that there is a limit on the maximum number of amplifier spans that can be concatenated. It is also not possible to monitor the quality of the individual channels at each amplifier site, but this is not a major problem, since the gain-flattened amplifiers essentially treat all channels the same. In this case, transparency wins over electrical regenerators, and that why it is widely used. E.1.2 Add-drop multiplexers Add/Drop Multiplexer is a potential element for optical transparency. At any one node, the majority of the channels are in the pass-through state, and thus using optically transparent passthrough obviates the need for a lot of OEO conversions. But there are still many technologies competing in this area with no obvious winner even if some transparent OADM is available. It is hard to make any projection and what will be really the advantage of transparency even if one can sense that transparent and reconfigurable Add/Drop will have some great advantages over opaque component if only for the cost of the signal conversion. For transparency to be adopted in this area, adequate solutions have to be found; only then transparent optical Add/Drop elements would begin to be widely rolled out in large optical networks. E.1.3 Wavelength conversion techniques Wavelength conversion is the technique of transferring the information modulated on an optical carrier to another optical carrier of a different wavelength. Such a conversion can allow wavelength reuse among several all-optical network sections by transferring data to any available wavelength. Wavelength conversion is particularly useful for routing or switching functions. There is less contention establishing a light-path with wavelength conversion than using the same wavelength from end to end. Wavelength conversion helps to reduce the WDM connection blocking probabilities. But optical wavelength conversion technology is still immature and remains expensive at present. Wavelength conversion is most useful at the edge of a network administrative domain to reduce the information correlation and dependencies across domains. E.1.4 Optical cross-connects Typical applications for cross-connects will be to manage provisioning, protection, and restoration in large scale networks. The key feature of the all-optical cross-connect is that it can pass signals in any data rate or format. However, it cannot easily monitor signal quality or identity, and it cannot provide drop & continue. In addition, it cannot remove signal impairments, or rewrite signal overhead data. Some of the problems with the all-optical cross-connect can be fixed by adding transceivers. In this case, the configuration is not totally independent of the data rate and format, but since the switch fabric is optical, one can upgrade individual paths by replacing the transceivers for those paths. No 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 88 (93) EURESCOM Technical Information changes would be necessary in the switch fabric or in the other transceivers. Since all signals pass through at least one transceiver, it is possible to monitor the signal quality and identity, to regenerate the data to remove the accumulated impairments, and even to rewrite signal overhead data. But, as in the totally transparent case, the switch fabric does not provide intrinsic capability for functions such as drop & continue. Cross-connects with electronic (opaque) switch fabrics provide all of the features of an optical fabric with transceivers, except that the fabric is not completely format or bit-rate independent. One major drawback is that this kind of cross-connect does not scale well. To summarise the transparency issues for cross-connects, it appears that totally transparent crossconnects will simply not be able to meet all the system requirements. But more and more vendors are designing cross-connects with transparent fabric with transceivers in order to make them more attractive. These cross-connect scale very well because to upgrade the bit-rate of any path, only the transceivers have to be changed, thus preserving the investment of the carrier. Another solution is to make hybrid designs combining optical and electronic fabrics in order to exploit their different advantages. But what is missing is the ability to translate one wavelength to another, to add/drop wavelength without regeneration, to read bits/byte/frames with photons, to handle the restoration on a per wavelength basis and to reach an adequate performance management. Research for all-optical switches will continue because it is driven by the expectation of significant savings in cost, power and space that can be achieved by eliminating O-E-O conversions and the underlying cost of the electronics. But it is seemingly clear that practical core networks will be deployed in opaque form. The merit of opaque cross-connects is that they are capable to arrest cumulative transmission impairments. Thus, they ease the management of the optical network. E.2 Conclusion As a technical solution, the use of optical transparency in the optical networks should provide cost efficient solution in terms of equipment cost. A typical argument for this is that compared to opaque cross-connects for instance, transparent nodes save transponders used for OEO conversions. Together with cost savings are associated space and power savings, reliability and so on. On the other hand, today’s use of transponders is not limited to the OEO and regeneration function. The use of transponders as network element interfaces including management functions points out the main issues of the transparent approach. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 89 (93) Annex F Requirements on the control plane of ASON F.1 Introduction A control plane architecture for optical transport network must be flexible and potentially applicable to all circuit switched networks be it at the fibre level (OTS), the waveband level (OMS) or the wavelength level (OCh). Unlike data networks, Automatically Switched Optical Networks (ASONs) have their own features that should be taken into consideration in the design of pertinent control technologies and architectures, hence the need to add extensions to the existing protocols. Some of these features are: Network level features: ASON serves as the backbone for user traffic. The traffic load and topology changes are not as dynamic as those for many packet switched data networks. Paths may belong to different groups: permanent, semi-permanent or fast provisioned (short-lived connections). ASON provides services with different types of CoS/SLAs (BER, availability, maximum downtime, delay etc.) than those that typically prevail for data networks. Fast restoration is of paramount importance for some services. ASON may have additional link state parameters than the link state metrics used for data connections. As path selection is strongly technology dependent, it is suggested that it is done within one domain. Circuits are set up with many technology dependent constraints. The path selection process for circuit connections can be impacted by technology-specific considerations and by the specific vendor design choices such as wavelength band/spacing, power levels, modulation methods and Forward Error Correction methods used. While it is true that the grid of usable wavelengths has been standardised, no agreement has been reached yet on which bands (with defined number of wavelengths) would be used by all manufacturers to have seamless overlap. Data paths (LSPs) are generally considered to be unidirectional while circuit connections are generally bi-directional. The lack of wavelength conversion could impact strongly the routing process because a common wavelength must be available on all the links used for the path. This problem needs further studies. But if wavelength conversion is available at each node, no information is needed to assess the availability of particular wavelengths on a link. Network element level features: Circuits switched network elements (NEs) may have hundreds or even thousands of physical ports. Many of them terminate on the same neighbour and share the same attributes. Ports in circuit switched NE are not typically assigned IP addresses (IP + local ID which could be vendor defined). Each physical port includes many technology dependent attributes. F.2 Routing Routing requires the availability of the view of the network topology that can be used for operations and end-to-end path communication. To discover the topology of the network, state information dissemination is therefore necessary. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 90 (93) EURESCOM Technical Information State information dissemination encompasses the manner in which NE level resource information is disseminated throughout the network to form network level resource database or link state database. State information distribution involves many steps. First, the NE level physical resource map is acquired and summarised in logical links based on link attributes according to a pertinent trade off in order to avoid a glut of information. They are then distributed throughout the network. Key issues that relate to state information dissemination include the specific types of state information that needs to be distributed and the mechanisms for representing and disseminating this information. Link-state routing protocols, such as IS-IS and OSPF, have been used for some time in the IP world to compute destination-based next hops for routes (without routing loops). In non-packet networks, however, their immense value comes from their ability to provide timely topology and network status information in a distributed manner, i.e., at any network node. But it is also important to be aware that optical layer routing is less insulated from details of physical implementation than routing in higher layers (IP). OSPF/IS-IS mechanisms with the appropriate extensions can be used to disseminate link state information within a single area or autonomous system. This approach to extend existing protocols minimises risk while reducing time to market as we are dealing with proven routing protocols. Link state information includes both static and dynamic information as follows: Static information is required for circuit operation. It includes neighbour connectivity, logical attributes, total bandwidth etc. This information does not need to be propagated very often as it changes less frequently. Some dynamic information is required for circuit operation, while others are mainly used for traffic engineering purposes. It includes bandwidth availability etc. This information is up-dated as frequently as possible. Topology updates should happen only when a significant topology change occurs or upon expiration of a periodic timer. Periodic topology refreshing is needed to ensure that network topology information in all NEs is consistent and correct. F.2.1 Resource discovery stages Resource discovery is identified as the transaction that establishes, verifies, updates and maintains the NE adjacencies and their port-pairs association for transport plane. Resource discovery can be achieved through either manual configuration (to be prohibited whenever possible because highly error-prone) or automated procedures. The resource discovery module generates a complete NE level resource map which includes attributes, neighbour identifiers and pertinent real-time operational status. The control procedure of this component (resource discovery function) should be very generic. At the same time, some of its sub-component may have to be technology and perhaps even product specific. Resource discovery may be extended to include service/policy negotiation between adjacent NEs. Typically the following steps are involved in resource discovery: Self-discovery: the results of self-discovery is to populate the local ID, physical attributes, logical constraints parameters in the network element resource table. Neighbour discovery and port association: This step discovers the adjacencies in the transport plane and their port association and populates the neighbour NE address and port ID fields in the resource table. Neighbour discovery protocols (NDP) are generally proprietary protocols with all the limitations inherent to non-standard protocols. Because the control plane network's topology may be different from the transport plane network's one in ASON, network elements that are not adjacent in the control plane may be adjacent in the transport plane. In order to unambiguously identify the transport plane neighbours and their port associations, it is essential to have in-band events (along the transport plane) coordinated with other EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information page 91 (93) control messages (along the control plane). Otherwise, the configuration has to be done manually. This point needs further studies as it is not yet clear how to achieve coordination between in-band and out-of-band messages in all circumstances (all-optical switches or totally transparent NEs). Transmitting control plane information separately from the data plane would surely increase the efficiency as the data could be switched transparently. The control channel could be transmitted along a separate wavelength or fibre or over an Ethernet link. Implementation of the control plane network is out of the scope of this document. F.2.2 Path selection Transport network routing procedures typically utilise explicit routing, where path selection can be done either by an operator through the use of simulation and planning tools, by a NMS or by the network itself. Hop by hop routing is also possible and may have its own benefits under some limited circumstances. Unlike packet switching, when setting up circuits in an optical transport network, we are dedicating bandwidth for the exclusive use of the two end systems involved with the connection. A path is normally requested with certain constraints which mainly come from the customers’ requirements, operators’ traffic engineering policies and network characteristics. Path selection request should use constrained-based routing algorithms that compute paths in the presence of multiple constraints and objectives not typically found in data networks: Conform to constraints such as physical diversity. Diversity is the possibility to route two paths between a single pair of points so no single route failure will disrupt them both; Achieve traffic engineering objectives in the transport network; Adhere to operator policies on routing such as preferred routes, excluded nodes, security etc.; Conform to network specific constraints. Path selection could be done either off-line or on-line. The path selection algorithms may also be executed in real-time or non real-time depending upon their computational complexity and specific network context. Off-line computation is typically a centralised operation whereas on-line computation is typically distributed. Explicit routing is preferred as it provides superior traffic engineering capabilities. Explicit routing may in fact be mandatory if multiple constraints are considered in the path selection decision because the rate of success, in this case, for path creation is much higher. Hop by hop can make more accurate next hop choice because each node can consult its local resource information that may be unavailable to other nodes. However, hop-by-hop routing requires path calculation at each node and needs loop-prevention mechanisms. Hop-by-hop routing is also more difficult when constraints other than additive link metrics, which is often the case for optical networks, are taken into account. Wavelength conversion is an issue in itself encountered in optical networks. Wavelength conversion is typically accomplished via use of electrical transponders. End-to-end connections employing the same wavelength through the network may pose considerable challenges in path selection. Advertising detailed wavelength availabilities on each link is not likely to scale. This problem is not solved yet and deserves further studies. While distributed, dynamic path selection has significant advantages such as scalability, it is strongly advisable to provide the operator some degree of control over path selection which comply with his own policy. F.2.3 Transactions The basic transaction for path management are: Create: this transaction creates a link of end-to-end path. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012 page 92 (93) EURESCOM Technical Information Delete : this transaction de-allocates or tears down a specific link of an LSP identified by its LSP ID. Query: this transaction queries the status of a specific link of an existing LSP identified by a LSP ID. F.3 Logical link/bundle Typically, optical switches may have hundreds or thousands of physical ports. Minimising global information and keeping information and decision making locally as much as possible is therefore necessary. Hence, link aggregation plays a key role in ASON topology representation. Parallel link connections can be summarised into an aggregate logical link, which hides the link connection details not necessary for network level control functions, such as path selection. The summarised information is then distributed to the rest of the network. The more details that are depicted in the topology database, the more accurate the selection could be; but too much information could result in an implementation that is not scalable. In general, scalability and simplicity should be favored above abundance of information. Stability is another crucial consideration. The more the amount of information conveyed and the more the network will tend towards less stable operating regimes (convergence issues). LMP(Link Multi-Protocol), which is under development at the IETF, is a protocol that runs between adjacent nodes and is designed to provide four basic functions for the node pair: control channel management, link connectivity verification, link property correlation and fault isolation. Control channel management is used to establish and maintain link connectivity between neighboring nodes. This is done using short Hello messages that act as a fast keep-alive mechanism between the nodes. Link connectivity verification is used to verify the connectivity of the links. Link property provides correlation function of link properties (link IDs, protection mechanisms and priorities) between adjacent nodes. Finally, LMP provides a mechanism to isolate link and channel failures in both opaque and transparent networks, independent of the data format. F.4 Protection It belongs to each operator to define the actual service in terms of priority, protection and restoration options. Protection is a collection of proactive techniques meant to restore failed connections. It also addresses the problem of survivability of the network after a failure. It is obvious that the protection assumes redundant network resources. Dedicated protection (1+1 ): it is the fastest but also the least cost effective. It requires a dedicated physically diverse protection path for each working path. Traffic is duplicated and carried simultaneously on both paths. This bandwidth can be considered wasted and that is why it is not always favoured. Its main advantage is that it can achieve very fast restoration. Dedicated protection (1:1): it is fast and more cost effective than 1+1 protection. It is possible to carry traffic with lower priority in the protection path. Shared protection (1:n): it is fast and more cost effective than dedicated protection. In this case, multiple paths could share the same reserved protection. It is possible to carry traffic with lower priority in the protection path. Survivability can still be enhanced while minimising the wastes of network bandwidth. This protection must be well defined with a set of priorities in order to avoid conflicts. Unprotected: no protection is provided after a failure is detected. EDIN 0200-1012 2003 EURESCOM Participants in Project P1012 EURESCOM Technical Information F.5 page 93 (93) Pre-emptable: the path can be pre-empted according to a set of priorities defined by the operator. Restoration One of the major services provided by a transport network is restoration. Restoration is a collection of reactive techniques used to restore failed connections. Restoration introduces several routing constraints. A major one is that of physically diverse routing. Restoration is often provided by either pre-computing or finding alternate routes for connections in real-time. More complete notions of diversity can be addressed by logical attributes such as shared risk link group (SRLG). SRLG do not change frequently, so this information is distributed less frequently. In order to address this constraint, path selection algorithms are needed that can find a diverse pair of routes. Time to reroute traffic is becoming shorter and shorter. What is the limit for restoration time has to be found. But one can agree that the larger the capacity of a circuit, the less time it can be tolerated to be down. A few ms are often suggested. A restoration time between 10 and 200 ms is therefore recommended. Circuit LSP restoration is very important for transport network. Circuit LSP restoration typically includes following the stages: F.6 Failure detection Failure notification Traffic restoration Signalling Signalling information must be exchanged between the optical nodes within the network as well between the optical network (ASON) and its clients via the UNI. To boot-start the system optical Network Elements (NEs) must be able to exchange control information. One way to support this is to adapt the optical control channel principle here and pre-configure a dedicated wavelength as supervisory channel for control traffic. The transparency and multi-protocol properties of this channel will allow ASON to accommodate many different clients. It is important to define what type of optical network information needs to be exchanged between nodes within ASON as well as the type of information that needs to be exchanged between ASON and its clients. Two signalling protocols, RSVP-TE and CR-LDP, are used in MPLS. With optical extensions, these protocols will be used in GMPLS. F.7 Control plane failure handling As the control plane and the transport plane are supposed to be separated, one way to handle control plane failures is to keep current circuit LSPs in operation but abort requests that are being processed. Control network failure should be capable to allow dissemination of failure events to a management system for management purposes. But to prevent such extreme events from happening, it is advisable to design robust control planes, with maybe the right backup, which could ensure survivability in the harshest situations. 2003 EURESCOM Participants in Project P1012 EDIN 0200-1012