Principle of ATM (2) 2001 Copyright SCUT DT&P Labs 1 5. ATM Adaptation Layer 2001 Copyright SCUT DT&P Labs 2 5. ATM Adaptation Layer 1. QoS Service Catagories CBR Constant Bit Rate VBR-RT Variable Bit Rate - Real Time VBR-NRT Variable Bit Rate - Non-Real Time ABR Available Bit Rate UBR Unspecified Bit Rate GFR Guaranteed Frame Rate (later) 2001 Copyright SCUT DT&P Labs 3 5. ATM Adaptation Layer 1. QoS Service Catagories (cont.) CBR has been defined to support constant bit rate connectionoriented traffic where end-to-end synchronisation is required. This is otherwise known as ITU-T Class A performance requirements. VBR-RT has been defined to support variable bit rate connection-oriented traffic where end-to-end synchronisation is required. This is otherwise known as Class B performance requirements. VBR-NRT is for types of traffic which are predictable, yet do not require a timing relationship to be maintained end-toend. ABR service is designed for economical support of applications with vague requirements for throughputs and delays. 2001 Copyright SCUT DT&P Labs 4 5. ATM Adaptation Layer 1. QoS Service Catagories (cont.) UBR operates on a 'best effort' basis, with no reservation of bandwidth. Signalling used to set up and clear down calls is normally transmitted as UBR,as is Local Area Network Emulation (LANE) traffic. GFR is a new service category which is still being defined. It is intended to provide a mechanism that will deliver frames (as cells). If one cell is lost they are all lost. What is guaranteed is a frame rate rather than a cell rate. 2001 Copyright SCUT DT&P Labs 5 5. ATM Adaptation Layer Classes as defined by ITU-T rec. I 362 Class A Timing between source and destination Bit rate Class C Required Constant Connection mode Relevant Adaptation Layer Class B Not required Variable Connection-oriented AAL 1 Class D AAL 2 Connectionless AAL 3 AAL 4 AAL 5 2001 Copyright SCUT DT&P Labs 6 5. ATM Adaptation Layer 2. General Principles of Adaptation The use of a CS is not required by all AALs Higher layer data Etc. CS T H H T H T Adaptation Layer SAR 2001 Copyright H SCUT DT&P Labs H 7 5. ATM Adaptation Layer 3. Usage of Adaptation Layer AAL is used to adapt a source application to ATM ATM switching takes place in the ATM Layer. 2001 Copyright SCUT DT&P Labs 8 5. ATM Adaptation Layer 4. AAL1 • A function of the AAL associated with Class A data, AAL1, is to ensure that there is timing integrity between the sending and the receiving end. • Another function is to carry out clock recovery at the destination. • The AAL also provides a mechanism to detect lost cells, and inserts a dummy into the cell stream to ensure that the timing information is not lost. 2001 Copyright SCUT DT&P Labs 9 5. ATM Adaptation Layer 4. AAL 1 (cont.) To format Class A data into cells, the data stream at the defined operating speed is simply chopped up into 47-byte chunks. Each 47-byte SDU is preceded by a one-byte header, resulting in a 48-byte payload. 2001 Copyright SCUT DT&P Labs 10 5. ATM Adaptation Layer 4. AAL 1 (cont.) The SN field is then split into two parts: the Convergence Sublayer Indication bit (CSI) which is normally set to 0, and three bits for the Sequence Number.This cycles through from 0 to 7 and back to 0 again, and is suitable for identifying missing or misinserted cells. To ensure the integrity of the SN field, it is protected by the SNP (Sequence Number Protection) field, which is a three-bit CRC check with an additional even-parity bit. 2001 Copyright SCUT DT&P Labs 11 5. ATM Adaptation Layer 5. AAL2 • AAL2 defines the transport of VBR traffic that is timingsensitive, such as VBR audio and video. • A feature of AAL2 is the ability to accept several streams of traffic and multiplex them together. 2001 Copyright SCUT DT&P Labs 12 5. ATM Adaptation Layer 5. AAL 2 (cont.) 2001 Copyright SCUT DT&P Labs 13 5. ATM Adaptation Layer 2001 Copyright SCUT DT&P Labs 14 5. ATM Adaptation Layer 5. AAL2 (cont.) Initial AAL 2 header • CID Field The channel identifier field identifies the individual user channels within the AAL2, and allows up to 248 individual users within each AAL2 structure. • LI Field The length identifier identifies the length of the packet payload associated with each individual user, and assures conveyance of the variable payload. • UUI Field One current use for the User-to-user field is to negotiate a larger Maximum Transfer Unit (MTU) size for IP. 2001 Copyright SCUT DT&P Labs 15 5. ATM Adaptation Layer 5. AAL2 (cont.) Secondary AAL 2 header The Offset Field identifies the location of the start of the next packet within the flow. For robustness the Start Field is protected from errors by the Parity bit (P) and data integrity is protected by the Sequence Number (SN). 2001 Copyright SCUT DT&P Labs 16 5. ATM Adaptation Layer 6. AAL3/4 • The two workgroups, AAL3 and AAL4,discovered that they had produced near-identical processes. The two work groups subsequently joined forces to produce the single adaptation known as AAL 3/4. • AAL 3/4 has a relatively high overhead. In this case, 4 octets are consumed by the header and trailer fields. After subtracting this overhead, the payload has been reduced to 44 octets. • Although originally designed to carry all manner of traditional data traffic, AAL 3/4 was seen as overly complex to implement and also as inefficient due to its high overheads. Consequently, most data traffic is carried in AAL 5. 2001 Copyright SCUT DT&P Labs 17 5. ATM Adaptation Layer 6. AAL 3/4 (cont.) 2001 Copyright SCUT DT&P Labs 18 5. ATM Adaptation Layer 6. AAL 3/4 (cont.) 2001 Copyright SCUT DT&P Labs 19 5. ATM Adaptation Layer 6.1 AAL3/4 CS • Type Indicates the units used by the BA and Length fields. • BTag/Etag: These two 'tags' are a numerical value (the same value), which help to ensure that it is a single CS unit that has been received and not a damaged CS unit created by joining together parts of two CS units. • BA Size Length of the user information subfield of the CS payload. • Pad Padding added to ensure that the total length of the CS is divisible by 4 (32 bits).This is an engineering consideration to simplify processing by 32-bit processors. • Length of the user information subfield. Other fields and subfields are reserved for future definition. 2001 Copyright SCUT DT&P Labs 20 5. ATM Adaptation Layer 6.2 AAL 3/4 SAR • Segment Type Indicates whether a cell is the first (at the beginning) of a message (BOM), a continuation of a message (COM), or the last (at the end) in a message (EOM). • Sequence Indicates the position in a convergence PDU of a SAR PDU. • MID: This is multiplexing ID field. This can be used to allow the multiplexing of several traffic streams into a single connection. • Len: This is the length of the actual data in the last cell of a message. • CRC: A 10-bit Cyclic Redundancy Check computed over the SAR PDU. 2001 Copyright SCUT DT&P Labs 21 5. ATM Adaptation Layer 7. AAL5 • AAL5 has significantly lower overheads than AAL 3/4 and is, therefore, very widely adopted. • In practice, since AAL 2 is not yet widely used and AAL 3/4 is seen as overly complex and cumbersome, only AAL1 and AAL5 are widely used. • AAL1 is used for CBR traffic and AAL5 for all others: VBR, UBR and ABR. 2001 Copyright SCUT DT&P Labs 22 5. ATM Adaptation Layer 7. AAL5 (cont.) 2001 Copyright SCUT DT&P Labs 23 5. ATM Adaptation Layer 7. AAL5 (cont.) AAL5 Frame Format • AAL5 simply takes the network layer packet and adds a single trailer. • The PAD field is there to pad out the complete PDU so that it can be divided into an integer number of 48-byte segments for loading into the cells. AAL5 Trailer The AAL5 8-byte trailer consists of: • Two 1-byte fields which are unused • A 2-byte length field which indicates the length of the data, not including the trailer and pad • A 4-byte CRC 2001 Copyright SCUT DT&P Labs 24 5. ATM Adaptation Layer 7. AAL5 (cont.) AAL5 Trailer 2001 Copyright SCUT DT&P Labs 25 5. ATM Adaptation Layer 7. AAL5 (cont.) AAL5 Transmission • The PTI field in the header is used. Bit 1 is set to 1 when the last cell representing the PDU is assembled, and all other cells have the bit set to 0. • When the receiver sees the PTI field with bit 1 set to 1, it assumes that the next cell with the same VPI/VCI number will be the first cell of a new PDU. 2001 Copyright SCUT DT&P Labs 26 5. ATM Adaptation Layer 7. AAL5 (cont.) AAL5 Transmission AAL5 makes use of the PTI field in ATM cell header Bit 1 = 1 indicates this cell carries the AAL5 trailer GFC VPI VPI VCI VCI VCI PTI CLP HEC 48-byte data field 2001 Copyright SCUT DT&P Labs 27 6. Signalling 2001 Copyright SCUT DT&P Labs 28 6. Signalling 1. Signalling Functions • If connections are to be set up on demand, a form of signalling is essential. Connections set up in this way are referred to as switched virtual circuits (SVCs). • it is necessary to adopt a signalling system which is internationally accepted together with an addressing scheme which operates on a global basis. • The ITU-T standard for signalling in ATM public networks is known as Q.2931. • The ATM Forum derived two separate standards from this for private networks, known as V3.0 and V3.1. • UNI 4.0 has also been released. This brings the ATMF signalling subset closer to Q.2931. 2001 Copyright SCUT DT&P Labs 29 6. Signalling 1. Signalling Functions (cont.) • All signalling within ATM is carried over a standard reserved channel: VPI=0, VCI=5. • Signalling is separately defined for use across the UNI (Q.2931, UNI 3.0, UNI3.1, UNI 4.0) and for use at the NNI for setting up the calls (PNNI). 2001 Copyright SCUT DT&P Labs 30 6. Signalling 2. Signalling Control Functions Establishing a virtual circuit Status report for a virtual circuit Maintaining a virtual circuit Clearing a virtual circuit 2001 Copyright SCUT DT&P Labs 31 6. Signalling 3. Address Formats • Work is still proceeding on defining the most effective addressing structures for use in ATM. Below are listed three formats that are used in private networks. • The carriers have already declared their intent to use E.164 addresses. 2001 Copyright SCUT DT&P Labs 32 6. Signalling 3. Address Formats (cont.) D.C.C. Format AFI High order DSP DCC IDI E.S.I. Sel. DSP IDP I.C.D. Format AFI High order DSP ICD IDI E.S.I. Sel. E.S.I. Sel. DSP IDP E.164 Format AFI E.164 ISDN Number High order DSP IDI DSP IDP 48-bit ‘MAC’ address 2001 Copyright SCUT DT&P Labs 33 6. Signalling • DCC ATM Address Format Authority Format Indicator (AFI) = 39 Data Country Code (DCC). • ICD ATM Address Authority Format Indicator (AFI) = 47 International Code Designator (ICD) • NSAP Encapsulated E. 164 Address Format Authority Format Indicator (AFI) = 45 E.164 - An E.164 format (telephone) number (NSAP: Network System Access Point) General Domain Specific Part (DSP) End System (or Station) Identifier (ESI) Sel Selector 2001 Copyright SCUT DT&P Labs 34 6. Signalling 4. Call Set-up • A call set-up message is sent by the calling party into the network to initiate a connection. • It is also passed from the network to the called party to initiate the connection. • Assuming successful call establishment, the called party will respond with a connect message. • With ATM we need to specify a list of characteristics that the network must support, for example, the quality of service (QoS). 2001 Copyright SCUT DT&P Labs 35 6. Signalling 4. Call Set-up (cont.) Calling Party UNI 2001 Copyright UNI SCUT DT&P Labs Called Party 36 6. Signalling 4. Call Release • The release message can be sent by either party to clear down the connection. • If one party clears, then the network will send a clear message to the other party. • The network may also initiate the clear-down if, for example, a network failure occurs, or in the absence of traffic for a pre-determined time period. 2001 Copyright SCUT DT&P Labs 37 6. Signalling 4. Call Release (cont.) Calling Party UNI 2001 Copyright UNI SCUT DT&P Labs Called Party 38 6. Signalling 5. Point-to-Multipoint Connections • Multipoint connections are a feature of ATM networks. They are used in all LAN techniques. • They will be a most important feature of broadcast networks such as those providing video on demand. • The process of setting up a point-to-multipoint connection involves first of all setting up a point-to-point connection. It must be specified that this connection is to be multipoint (This must be done as multipoints are uni-directional.单向 的) 2001 Copyright SCUT DT&P Labs 39 6. Signalling 5. Point-to-Multipoint Connections • Once the initial point-to-point is set up additional destinations (leaves) can be added. There are two alternative mechanisms that can be used here: (1) Send a request to the root (the originator of the original point-to-point); (2) With signalling version 4.0 issue a Leaf Initiated Join (LIJ) request to the network. 2001 Copyright SCUT DT&P Labs 40 6. Signalling 5. Point-to-Multipoint Connections (cont.) ROOT Party UNI UNI NEW Leaf Point-to-Point Connection Point-to-Multipoint Connection 2001 Copyright SCUT DT&P Labs 41 6. Signalling 6. The Traffic Contract • The traffic contract is the sum total of all the parameters required to define the characteristics of a connection. • The contract includes an indication of how the network is to verify that the user does not use more resources than were requested at set-up time. • The contract consists of a series of requirements that are encoded for transmission to the network at the ingress switch to the network (this includes a value of required bandwidth and delay). 2001 Copyright SCUT DT&P Labs 42 6. Signalling 6. The Traffic Contract (cont.) • The set-up message carries the destination 20-byte ATM addresses, plus the basic bandwidth parameters forward and reverse, and the QoS class. • The set-up message may also carry the source ATM address. • The traffic contract between user and network establishes: – Virtual bandwidth reserved in each of the forward and backward directions; – QoS class for cells in each of the forward and reverse directions. 2001 Copyright SCUT DT&P Labs 43 6. Signalling 6. The Traffic Contract (cont.) • The Connection Admission Control (CAC) algorithm of the switch will then assess the network in the light of the request, before allowing the connection to proceed to setup. • The ingress switch will retain a copy of the pertinent parameters (such as PCR,SCR and MBS) and will use this information to check that the connection stays within its contracted bounds (a policing function). 2001 Copyright SCUT DT&P Labs 44 6. User-Network Interface (UNI) Signalling 2001 Copyright SCUT DT&P Labs 45 6. UNI Signalling 1. The User-Network Interface • The user-network Interface (UNI) is that point between the end-point ATM equipment and the first ATM switch. • There have been several versions of the UNI specification, defined by the ATM forum: UNI 2.0, UNI 3.0, UNI 3.1 and UNI 4.0 (also known as Sig 4.0). • Of these specifications UNI 2.0 supports only PVCs, while the latter three versions also support SVCs. • The ATMF signalling (from UNI 3.1 onwards) was aligned with the ITU-T Q.2931 signalling standard. 2001 Copyright SCUT DT&P Labs 46 6. UNI Signalling 1. The User-Network Interface (cont.) The set-up message will carry the source and destination ATM addresses, plus the bandwidth and the QoS parameters The call set-up message is chopped up using AAL5 and sent on reserved channel (VPI= 0, VCI=5). 2001 Copyright SCUT DT&P Labs 47 6. UNI Signalling 2. Q.2931 Signalling Format • Signalling under ATM consists of joining together a variety of basic building blocks containing the necessary information. • These building blocks are known as Information Elements (IEs) and each element has a standard 4-byte header followed by the IE content. • IEs are built as required by the message type and service type. 2001 Copyright SCUT DT&P Labs 48 6. UNI Signalling 2. Q.2931 Signalling Format Message Types: Call Establishment: CALL PROCEEDING CONNECT CONNECT ACKNOWLEDGE SETUP Call Clearing: RELEASE RELEASE COMPLETE RESTART RESTART ACKNOWLEDGE 2001 Copyright SCUT DT&P Labs 49 6. UNI Signalling 2. Q.2931 Signalling Format Message Types: Miscellaneous: STATUS STATUS ENQUIRY Point-to-Multipoint: ADD PARTY ADD PARTY ACKNOWLEDGE ADD PARTY REJECT DROP PARTY DROP PARTY ACKNOWLEDGE 2001 Copyright SCUT DT&P Labs 50 7. Private Network-to-Network Interface (PNNI) 2001 Copyright SCUT DT&P Labs 51 7. PNNI 1. PNNI Overview • Although the PNNI specification has been issued for use in private networks, PNNI proves to be sufficiently scalable and robust to be used in public networks. • It is likely that the official NNI signalling standard, when it is eventually released, will be strongly based on PNNI. 2001 Copyright SCUT DT&P Labs 52 1. PNNI Overview (cont.) 7. PNNI Signalling protocol to set up connections based on routing information Routing protocol to distribute reachability, capacity and QoS information 2001 Copyright SCUT DT&P Labs 53 7. PNNI 2. PNNI Targets To distribute among all participating switches the topology of the ATM network To operate at the network-network interface To allow for scalability by the creation of groups of switches To allow switches to build routing tables from the topological information To allow for ‘crank back’ to last the confirmed point and a search for an alternate route On end-to-end route confirmation, interface with Connection Admission Control to accept ATM call set-up 2001 Copyright SCUT DT&P Labs 54 7. PNNI 2. PNNI Targets (cont.) PNNI Standards • PNNI is an interface specification that uses a Link State process for the distribution of routing information. • The ATM Forum standard is P-NNI version 1.0 af-pnni0055.000 March 1996. • Error corrections issued as af-pnni-0081.000 July 1997. • PNNI supersedes an earlier version from December 1994 called Interim Inter Switch Protocol (IISP). This is sometimes referred to as PNNI phase 0. 2001 Copyright SCUT DT&P Labs 55 7. PNNI 3. PNNI Base Level • All ATM switch ports have an ATM address of 20 bytes (1) The first 13 bytes are normally fixed for each switch (2) The final 7 bytes are the physical address of the attached device (6 bytes, referred to as the MAC address) plus the selector field which is one byte. • In the example used here, we have only shown the last few hex digits of an ATM address. 2001 Copyright SCUT DT&P Labs 56 7. PNNI 3. PNNI Base Level (cont.) 2001 Copyright SCUT DT&P Labs 57 7. PNNI 3. PNNI Base Level (cont.) Forming Groups • On start-up, PNNI nodes send 'hello' packets on all interfaces to discover neighbours. • As part of this process, neighbouring nodes exchange their ID numbers. In this example, a 13-digit match is required. • All nodes with matching numbers form a logical peer group using the matching digits as a group identifier, for example, group number 202. • Nodes with at least one link terminating at a switch in a 'foreign' group are considered 'Border Nodes'. 2001 Copyright SCUT DT&P Labs 58 7. PNNI 3. PNNI Base Level (cont.) Information Exchange • Nodes within a group exchange and relay information about link status including virtual bandwidth, availability and next hop. • A reliable transport mechanism is used to ensure that all nodes ultimately share the same database. 2001 Copyright SCUT DT&P Labs 59 3. PNNI Base Level (cont.) 7. PNNI PNNI Groups 2001 Copyright SCUT DT&P Labs 60 7. PNNI 4. Peer Group Leader • As part of this process the nodes within a group select a group leader based on a configured priority number, or by selecting the node with the lowest address. • Group leaders establish logical connections with each other and exchange a summary of information about their groups. 2001 Copyright SCUT DT&P Labs 61 4. PNNI Peer Group Leaders (cont.) 2001 Copyright SCUT DT&P Labs 7. PNNI 62 7. PNNI 4. PNNI Logical Network • The network, viewed from the perspective of a group leader, will appear to have the topology as shown in the next diagram. • Locally, a group leader will retain the detailed view of its own group including border nodes and therefore 'real' links to neighbouring groups. • Group leaders pass this logical network map to the members of their own group. Each PNNI node, therefore, has a detailed description of its own group and a logical map on how to get to any other group. 2001 Copyright SCUT DT&P Labs 63 7. PNNI PNNI Logical Network 2001 Copyright SCUT DT&P Labs 64 7. PNNI 5. PNNI Operation • When a UNI signalling request comes in from the endstation on reserved channel 0,5 its contents will be analysed within the switch. • The switch first performs a Connection Admission Control (CAC) algorithm which determines whether or not the switch has the resources necessary to handle the incoming call. • A Generic Connection Admission Control (GCAC) is then performed. 2001 Copyright SCUT DT&P Labs 65 7. PNNI 5. PNNI Operation • This GCAC algorithm determines whether or not the switches between the source and destination can handle the call. • Following the GCAC algorithm, the switch prepares a Designated Transit List (DTL) which is an entire route through the network for the signalling request. • This DTL is added onto the call set-up message and sent to the next node along the intended signalling route. 2001 Copyright SCUT DT&P Labs 66 Global topology as seen by node 10126 2001 Copyright SCUT DT&P Labs 7. PNNI 67 PNNI Designated Transit List 2001 Copyright SCUT DT&P Labs 7. PNNI 68 7. PNNI 6. PNNI Packets Standard Header There are many types of packet used for the PNNI protocol. However, they all start with a standard layout header. PNNI Signalling The format of PNNI signalling packets is based on UNI 4.0 (Q.2931) with additions to cater for the transit lists and crankback. 2001 Copyright SCUT DT&P Labs 69 7. PNNI 6. PNNI Packets (cont.) All PNNI packets have a common header: 0 2 Bytes Packet Type 4 Bytes Length Version Packet Type: 1 = Hello 2 = PTSP (PNNI Topology State Packet) 3 = PTSP Acknowledge 4 = Database Summary 5 = PTSE Request 2001 Copyright 5 Bytes SCUT DT&P Labs 6 Bytes Supported Most recently supported protocol version, used to align protocol versions between different nodes 70 8. Network Management 2001 Copyright SCUT DT&P Labs 71 8. Network Management 1. General Management Model The management protocols used are typically the Simple Network Management Protocol (SNMP) and, in more recent devices, the Hypertext Transfer Protocol (HTTP). Local Management With local management, the management terminal is plugged directly into the ATM switch, typically into an Ethernet port or a serial port. Initially an ATM switch may be configured in this manner. 2001 Copyright SCUT DT&P Labs 72 8. Network Management 1. General Management Model (cont.) In band Management • In managing an ATM network, management traffic is sent in band that is over the ATM network itself on an ATM connection. • An SVC, or more typically a PVC, may be used for this purpose. ILMI • The Interim Local Management Interface Protocol (ILMI) is a standard ATM management protocol. • ILMI works only across the UNI interface, that is between the end ATM station and the first ATM switch. 2001 Copyright SCUT DT&P Labs 73 General Management Model 2001 Copyright SCUT DT&P Labs 8. Network Management 74 8. Network Management 2. Interim Local Management Interface (ILMI) • The ILMI standards produced by the ATM Forum enable a number of network management functions to be performed across the UNI. • ILMI is a model based on the use of SNMP for the interchange of data, which is adapted via AAL5 and transmitted over a predefined VPI/VCI=0/16. • The important difference between SNMP, as discussed in the previous section, and ILMI is that ILMI is SNMP over AAL5 directly, that is, without using IP. • ILMI is positioned at the public and private UNIs. ILMI also runs between the public and private network. • If one wishes to access this information remotely then one must run a management agent locally to access the local MIB. 2001 Copyright SCUT DT&P Labs 75 8. Network Management SNMP and ILMI 2001 Copyright SCUT DT&P Labs 76 8. Network Management 3. ILMI Functions • ILMI was intended originally to handle only the address registration and de-registration process for each end station in an ATM network. • Now its functions have grown and now include many other housekeeping operations, including control information, switch configuration details, statistics relating to the ATM connections, and the physical and ATM layer data. 2001 Copyright SCUT DT&P Labs 77 ILMI Basic Requirements 8. Network Management Single MIB for each ATM device A Management Information Base (MIB), which contains data relating to the status of each end station or intermediate switch, is set up for the ATM system. ILMI provides: Status information Configuration information Control information ILMI handles: Address registration Address de-registration Switch configuration 2001 Copyright SCUT DT&P Labs 78 8. Network Management 4. ILMI Agents • The ILMI requires a management entity at each end of the interface. • The UNI Management Entity (IME) acts as the server to the network. • management station client, and performs all necessary communications tasks via AAL5 and ATM cells. • The IME also controls access to the MIB. 2001 Copyright SCUT DT&P Labs 79 8. Network Management 4. ILMI Agents (cont.) The Interface Management Entity (IME) in each ATM device does the following: Handles the communication Provides access to the MIB Co-ordinates between ATM and physical layer information There are two types of Interface Managed Entity (IME): uIME - User IME (ATM end device) nIME - Network IME (ATM switch) nIME uIME ILMI (SNMP over AAL5) Private switch Private UNI 2001 Copyright SCUT DT&P Labs 80 8. Network Management 5. ILMI ATM UNI MIB Tree ATM UNI ILMI MIB Physical Layer ATM Layer Virtual Virtual Path Channel Connection Connection ATM Layer Statistics Network Prefix Address Interface Index Interface Interface Index + Index + Prefix Address Common Specific Interface Index 2001 Copyright Interface Index Interface Index + VPI Interface Index + VPI + VCI SCUT DT&P Labs 81 8. Network Management 6. Address Registration • All ATM addresses consist of 20 bytes of data, made up of two distinct parts: The end-station address, which is 6 bytes of MAC data plus a selector byte; The network prefix, which is 13 bytes of data. • The registration of end-station devices is carried out by ILMI using a cold start trap from either the end-station or the ATM switch. 2001 Copyright SCUT DT&P Labs 82 8. Network Management 6. Address Registration (cont.) Operation • When the ATM interface in the end station is enabled, a cold start trap is transmitted out along VCI 16. • The ATM switch receives this start trap and replies with the prefix associated with that ATM switch. • The end station then adds its own MAC address and selector field to the prefix to form a full ATM address. • This address is sent to the switch where it is registered. 2001 Copyright SCUT DT&P Labs 83 8. Network Management 6. Address Registration (cont.) Cold start trap nIME uIME ILMI SNMP set message Network Prefix ILMI SNMP Response ACK / NACK ILMI SNMP set message Host Address (Prefix + MAC + Selector ) ILMI SNMP Response ACK / NACK SNMP in AAL5 on VCI 16 2001 Copyright SCUT DT&P Labs 84 8. Network Management 7. Device Check • uIMEs are declared ‘down’ if they do not respond after four consecutive polls. • A uIME is de-registered after the nIME declares that the uIME is down. • The uIME is de-registered by removing its address entry from the nIME address table. 2001 Copyright SCUT DT&P Labs 85 8. Network Management 7. Device Check (cont.) uIME nIME ILMI SNMP get message connectivity poll ILMI SNMP response message connectivity ACK 2001 Copyright SCUT DT&P Labs 86 9. ATM Traffic Descriptors 2001 Copyright SCUT DT&P Labs 87 9. ATM Traffic Descriptors 1. Traffic Management Traffic Management is essential for the proper operation of ATM. The aim is to ensure that all the different classes of traffic receive the appropriate handling. Main features of traffic management: Traffic Contract Connection Admission Control Traffic Shaping Traffic Policing 2001 Copyright SCUT DT&P Labs 88 9. ATM Traffic Descriptors 2. Traffic Descriptor Parameters These are the parameters requested at connection set-up time: Peak Cell Rate (PCR) Sustainable Cell Rate (SCR) Maximum Burst Size (MBS) Minimum Cell Rate (MCR) Cell Delay Variation Tolerance (CDVT) 2001 Copyright SCUT DT&P Labs 89 9. ATM Traffic Descriptors 3. Required Parameters for each Service Category Service Category PCR SCR MCR CDVT CBR VBR-RT VBR-NRT ABR GFR UBR 2001 Copyright SCUT DT&P Labs 90 9. ATM Traffic Descriptors 4. Peak Cell Rate Peak Cell Rate (PCR) is the absolute maximum rate at which the network guarantees cell delivery A user may send cells at this rate for a short period of time Rate is reduced to maintain an average (SCR) PCR is used by CBR, VBR and ABR service categories 2001 Copyright SCUT DT&P Labs 91 9. ATM Traffic Descriptors 5. Sustainable Cell rate Sustainable Cell Rate (SCR) is the average rate that a network guarantees cell delivery Users may burst above the SCR to the PCR (up to a maximum of BT) as long as they reduce their rate of flow to maintain this rate SCR is only used by the VBR QoS category 2001 Copyright SCUT DT&P Labs 92 9. ATM Traffic Descriptors 6. Burst Tolerance Burst Tolerance (BT) is the maximum time that the network will accept cell rates of PCR BT is only used by the VBR QoS category 2001 Copyright SCUT DT&P Labs 93 9. ATM Traffic Descriptors 7. Minimum Cell Rate Minimum Cell Rate (MCR) is the highest rate at which the network guarantees delivery of cells A user may attempt to send at higher rates at the risk of losing cells This parameter is used to support an ABR service 2001 Copyright SCUT DT&P Labs 94 8. Cell Delay Variation and Tolerance Variation in Cell Delay is a fact of life Delays are caused by : Multiplexing Queuing OAM cell insertion Physical Layer overhead An application may need a guaranteed limit on the degree of variation, a specified tolerance 2001 Copyright SCUT DT&P Labs 95 9. Quality of Service Parameters 2001 Copyright SCUT DT&P Labs 96 9. Quality of Service Parameters 1. Quality of Service Parameters Negotiable QoS Parameters Cell Loss Ratio (CLR) Maximum Cell Transfer Delay (Max CTD) Peak to peak Cell Delay Variation (peak-to-peak CDV) Non-negotiable QoS Parameters Cell Error Ratio (CER) Severely Errored Cell Block Ratio (SECBR) Cell Misinsertion Rate (CMR) 2001 Copyright SCUT DT&P Labs 97 9. Quality of Service Parameters 2. Cell Loss Ratio Ratio of cells successfully delivered to cells presented per VPI/VCI Lost cells CLR Total transmitted cells Cells in Cells out 2001 Copyright SCUT DT&P Labs 98 9. Quality of Service Parameters 3. Maximum Cell Transfer Delay Cell Transfer Delay (CTD) is the time a cell takes to traverse the network CTD is made up of Propagation Delay Transmission Delay Switching Delay Queuing Delay The Maximum Cell Transfer Delay (maxCTD) is the maximum allowable CTD 2001 Copyright CDT = time SCUT DT&P Labs 99 9. Quality of Service Parameters 4. Peak-to-Peak Cell Delay Variation Cell Delay Variation (CDV) CDV is a measure of the difference between actual time of delivery of a cell and expected time CDV highlights bursts of cells typical of LAN-generated traffic Without delay variation Cells in Real CDV 2001 Copyright SCUT DT&P Labs 100 9. Quality of Service Parameters 5. MaxCTD, Peak-to-Peak CDV and CLR Cell Arrival Distribution Fixed Transit Delay 2001 Copyright The CLR requested at connection setup time actually places a limit on the value of the percentage of the cell arrival probability distribution lying outside the maxCTD arrival times. Cells delivered late peak-topeak CDV maxCTD SCUT DT&P Labs 101 9. Quality of Service Parameters 6. Accumulation of QoS Parameters CDVs + maxCTDs CDV + maxCTD CDVs1 + maxCTD + CDVs2 + maxCTD + CDVs3 + = P2P-CDV maxCTD = Total maxCTD The maxCTD and CDV parameters passed with signalling SETUP calls are accumulated as the call progresses through the network. 2001 Copyright SCUT DT&P Labs 102 9. Quality of Service Parameters 7. One Point CDV Reference Cell Stream Actual Cell Stream RC0 AC0 RC1 AC1 RC2 AC2 AC3 RC3 RC4 AC4 RC5 AC5 CDV Cell Delay Variation = Reference arrival time - Actual arrival time = RCn - ACn Negative values = gaps in cell stream Positive values = cell “clumping” 2001 Copyright SCUT DT&P Labs 103 9. Quality of Service Parameters 8. Non-negotiable QoS Parameters Cell Error Ratio, where : Errored cells CER Successful ly transf ered cells Errored cells Severely Errored Cell Block Ratio, where : Severely errored cell blocks SECBR Total transmi tted cell blocks Cell Misinsertion Rate, where : Misinserte d cells CMR Time interval 2001 Copyright SCUT DT&P Labs 104 9. Quality of Service Parameters 9. Factors affecting QoS Parameters CDV CTD Propagation Delay Media Errors Switch Design Buffer Capacity Traffic Load Number of Nodes Network Failures 2001 Copyright CLR CER SCUT DT&P Labs CMR SECBR 105 9. Quality of Service Parameters 10. Required Traffic Descriptors and QoS Parameters 2001 Copyright SCUT DT&P Labs 106 10. Traffic Control 2001 Copyright SCUT DT&P Labs 107 1. Connection Admission Control Included in the process of establishing a virtual connection (VPI/VCI): Traffic descriptors are included either in UNI signalling (SVC) or management setup (PVC) Network switches check the traffic requirements against handling ability Admission of the connection is rejected if the network cannot guarantee that it will meet the requirements 2001 Copyright SCUT DT&P Labs 108 1. Connection Admission Control – CAC (cont.) NNI signalling Attempt to find a path to B Able to currently support this request Inform CAC of result UNI signalling I wish to connect to B with these QoS parameters CAC UNI signalling Call proceeding B 2001 Copyright SCUT DT&P Labs 109 2. Virtual Bandwidth Traffic descriptors such as CDV, PCR may be summarised as a connection with a specific virtual bandwidth (Vbw) requirement Switches along the intended path of the connection check for Vbw If a switch does not have sufficient Vbw: A new route is selected The request is denied 2001 Copyright SCUT DT&P Labs 110 2. Virtual Bandwidth (cont.) A separate Virtual Bandwidth algorithm is used for each Quality of Service Category. Service Category Bandwidth Allocated CBR VBR-NRT VBR-RT ABR UBR GFR PCR <= Vbw <= Link rate SCR <= Vbw <= PCR SCR <= Vbw <= PCR Vbw = MCR Under Review Vbw = MCR 2001 Copyright SCUT DT&P Labs 111 3. Traffic Shaping Traffic shaping is the name given to any technique at the user site to ensure that outgoing cells conform to the traffic contract This makes the customer premises equipment (CPE) a well-behaved user Throttling back to agreed rates can increase throughput as the need to retransmit network discarded cells is removed 2001 Copyright SCUT DT&P Labs 112 4. Traffic Policing A traffic contract exists across the UNI when a call’s descriptors are accepted Not all user devices will be well-behaved Traffic policing is necessary to ensure that badlybehaved devices do not interfere with other users Cells outside the limits of the contract will: Be discarded Have CLP set for discard at busy switches 2001 Copyright SCUT DT&P Labs 113 5. Generic Cell Rate Algorithm Policing traffic is performed by applying the Generic Cell Rate Algorithm (GCRA) GCRA is a continuous state ‘leaky bucket’ algorithm It checks that cell streams conform to PCR, CDVT, SCR and BT PCR & SCR require separate instances of the leaky bucket, hence switches employ a ‘dual state leaky bucket’ 2001 Copyright SCUT DT&P Labs 114 6. Leaky Bucket Algorithm cells Peak Cell Rate Bucket Tolerance factor to account for jitter - caller the Cell Delay Variation Tolerance (CDVT), measured in ms Cells which overflow will be dropped or have their CLP set to 1 2001 Copyright SCUT DT&P Labs 115 The End of Part 2 2001 Copyright SCUT DT&P Labs 116