"Deeply Programmable Network" Emerging Technologies for Network Virtualization and Software Defined Network (SDN) Aki Nakao The University of Tokyo Keynote Speech ITU-T Kaleidoscope Conference in Kyoto 2013/4/23 The University of Tokyo became an academic member of ITU-T in 2012 2 Interdisciplinary Initiative in Information Studies (III) • The latest graduate school of UTokyo established in 2000 • Multi-disciplinary graduate school including 50+ professors http://www.iii.u-tokyo.ac.jp/en/about.php#gakufu 3 5 Courses in the III Graduate School • • • • • Socio-information and Communication Studies Cultural and Human Information Studies I am in here conducting Applied Computer Science research on International Master s Doctoral Degree computer Program: Information, Technology, and networks Emerging Design and Informatics Society in Asia (ITASIA) 4 Future Network Research Proliferating... • • • • • Future Internet Architecture (FIA) in U.S. Global Initiative in Network Innovations (GENI) in U.S. Framework Programme 7 (FP7) in EU Horizon 2020 (2014-) in EU New Generation Network (NwGN) in Japan One of the most active research areas for building sustainable communities through new ICT Infrastructures 5 Roadmap of Future Network Standardization ‘06 ‘07 ‘08 ‘09 ‘10 ‘11 ‘12 ‘13 NWGN A R&D Start B C D A: 1St ITU-T Kaleidoscope Conference B: ITU-T FG-FN Start (SG 13 WG5 Q.21) C: Y.3001 (First ITU-T Recommendation for Future Networks ) D: Y.3011 (Framework of Network Virtualization for Future Networks) Slides from Prof. Tomonori Aoyama 6 First ITU-­‐T Kaleidoscope Academic Conference A New Genera;on Network -­‐ Beyond NGN -­‐ May 12, 2008 Tomonori Aoyama Keio University NICT Slides from Prof. Tomonori Aoyama 7 Slides from Prof. Tomonori Aoyama 8 Why Future Network is needed? Is the Internet of today full of problems? 9 Traffic Increase in Data Center Access http://www.internetevolution.com/author.asp?section id=715&doc id=175123. 10 World-Wide Mobile Traffic Volume Rapid Increase in Mobile Data Traffic Ericsson Japan 「Traffic and Market Report」(2012/6) 11 SPAM: The Most Annoying Problem of Today • 95% of all email traffic • As of August 2007, one in every 87 emails was a phishing attack • Targeted attacks on rise 20k-30k unique phishing Source: APWG (Nick Feamster s Talk @UTokyo) 12 Top-10 Spamming BotNets BotNet Size # of Spams Grum 600K 40B Bobax 100K 27B Pushdo 1.5M 19B Rustock 2M 17B Bagle 500K 14B Mega-D 50K 11B Maazben 300K 2.5B Xarvester 60K 2.5B Donbot 100K 800M http://www.techrepublic.com/ 13 • • Mobile Botnet Android phones are exploited as of 2012 (10k-30k devices) RootSmart backdoor (remotely controlling Android phones) Number of malware modifications for Android OS http://www.securelist.com/ 14 How can we resolve newly observed, constantly arising problems in the current Internet? In order to establish sustainable communities... 15 Paradigm Shift in Programming Model The Internet Cloud Data Center 1 Cloud Data Center 2 Open Programmability at Edges Future Network Cloud Data Center 1 Cloud Data Center 2 Open Programmability within Network Multiple Isolated Programming Environments (Slices) 16 Multiple Flexibly & Deeply Programmable Networks Slice 1 Cache Oriented NW Slice 2 Sensor Processing NW Slice N Cloud Access NW ID/Loc NW Content Oriented NW TV Broadcast NW Legacy NW Future NW “Slices” accommodate diverse networks Slice = a set of resources reserved across mul7ple network domains cloud1 Smartphones cloud2 Network Virtualization Platform 17 Multiple Flexibly & Deeply Programmable Networks Slice 1 Cache Oriented NW Slice 2 Sensor Processing NW Slice N Cloud Access NW ID/Loc NW Content Oriented NW TV Broadcast NW Legacy NW Future NW “Slices” accommodate diverse networks Slice = a set of resources reserved across mul7ple network domains cloud1 Smartphones cloud2 Network Virtualization Platform Objective: Enabling future network architecture that can be quickly and flexibly programmed to adapt to diverse demands 17 Three Emerging Areas of Study in Future Network Research • Network Virtualization (NV) • Software Defined Network (SDN) • Network Functions Virtualization (NFV) • And more? Deep Programmability in Network is a key... 18 Deep Programmability Problem: Inflexible Infrastructure cannot deal with constantly arising problems Solution: Flexible / Adaptive Communication Infrastructure Deep Programmability enables new network functionalities that can resolve the problems Supporting multiple isolated new functionalities flexibly and dynamically Data Plane Control Plane 19 Network Virtualization Programmability(NFV) Programmability (SDN) Network Virtualization In computing, network virtualization is the process of combining hardware and software network resources and network functionality into a single, software-based administrative entity, a virtual network. Network virtualization involves platform virtualization, often combined with resource virtualization. http://en.wikipedia.org/wiki/Network_virtualization The advanced network virtualization applies the virtualization technology to the networks in a sense that the networks offer extra capabilities provided by software resources in the nodes. These capabilities are beyond the basic network capabilities, i.e., link and their selection. The extra involvement enables the nodes, and the network, to provide computation and data storage capabilities to the users. The ICT infrastructure in the new-generation networks, which install the advanced network virtualization, govern a collection of the resources ranging from links, networks, and node-software as a slice and create a virtualized network over the slice with dynamically controllable and programmable links and nodes. http://nvlab.nakao-lab.org/nv-study-group-white-paper.v1.0.pdf 20 SDN ? 21 SDN • ? Software Driven Network (IETF BoF) 21 SDN • • ? Software Driven Network (IETF BoF) Software Defined Network 21 SDN • • • ? Software Driven Network (IETF BoF) Software Defined Network Some Definition Needed :-) 21 SDN • • • ? Software Driven Network (IETF BoF) Software Defined Network Some Definition Needed :-) Software-defined networking (SDN) is an approach to building computer networks that separates and abstracts elements of these systems SDN decouples the system that makes decisions about where traffic is sent (the control plane) from the underlying system that forwards traffic to the selected destination (the data plane)... ...this technology simplifies networking and enables new applications, such as network virtualization... http://en.wikipedia.org/wiki/Software-defined_networking 21 Standardization Activities • • ITU-T on Future Networks / Network Virtualization • • • IETF BoF on Software Driven Networks (SDN) ETSI on Network Functions Virtualization (NFV) IRTF BoF on Network Virtualization ONF (Open Networking Foundation) on Software Defined Networks (SDN) /OpenFlow • OpenDaylight on Software Defined Networks (SDN) 22 Y.3001: Future networks: Objec7ves and design goals • Defini7on of FN: – A network able to provide services, capabili7es, and facili7es difficult to provide using exis7ng network technologies. • 4 Objec7ves of FNs: – Service awareness : to provide services whose func7ons are designed to be appropriate to the needs of applica7ons and users – Data awareness : to have architecture op7mized to handling enormous amounts of data in a distributed environment, and are recommended to enable users to access desired data safely, easily, quickly, and accurately, regardless of their loca7on – Environmental awareness : to be environmentally friendly – Social and economic awareness : to consider social and economic issues to reduce barriers to entry for the various actors involved in the network ecosystem Service Service Diversity awareness These objectives are not considered as primary or not realized to a satisfactory extent in current networks. These objectives are the candidate characteristics that clearly differentiate FNs Target Date: roughly 2015-2020 Functional Flexibility Virtualization of Resources Data awareness Data Access Network Management Identification Mobility Reliability and Security Energy Consumption Optimization Environmental awareness Service Universalization Economic Incentives Social and economic awareness (Slides from Nozomu Nishinaga and Takashi Egawa). 23 Y.3011: Framework of network virtualiza7on for future networks § Definition of “network virtualization": Ø A technology that enables the creation of logically isolated network partitions over shared physical networks so that heterogeneous collection of multiple virtual networks can simultaneously coexist over the shared networks. This includes the aggregation of multiple resources in a provider and appearing as a single resource. § Motivation Various Services Coexistence of multiple networks Simplified access to resources Flexibility in provisioning Evolvability Design goals LINP2 Virtual Resources Physical NW 4 Manager Physical NW 4 Physical NW 1 Manager Physical NW 2 Physical NW 3 Physical NW 3 Manager Physical NW 1 Physical NW 2 Manager § Isolation, network abstraction, topology awareness and quick reconfigurability, performance, programmability, management, mobility, wireless Virtual Networks LINP1 Virtual Resources Manager Ø Ø Ø Ø Ø LINP3 LINP2 Manager § Problem spaces & design goals LINP1 Manager § Diverse services § Heterogeneous network architectures LINP3 Manager Ø Key technology for Service Awareness of FNs (Slides from Nozomu Nishinaga and Takashi Egawa). 24 Network Functions Virtualisation • Replaces proprietary hardware with software equivalents running on high volume industry standard servers, and using IT virtualisation techniques • Enables a wide variety of eco-systems and encourages openness. • Provides opportunities for pure software players. • Facilitates innovation towards new network functions and services that are only practicable in a purely software network environment. • Applicable to any data plane packet processing and control plane functions, in fixed or mobile networks • Enables new ways to implement resilience, service assurance, test and diagnostics and security surveillance • But Network Functions Virtualisation will only scale if management and configuration of functions can be automated • NFV aims to l ultimately transform the way network operators architect and operate their networks, but change can be incremental ETSI NFV "Network Functions Virtualisation", NFV ETSI Industry Specification Group, OFC - NFOEC, Mar.17-21, 2013 25 Network Functions Virtualisation: Vision Classical Network Appliance Approach Network Func7ons Virtualisa7on Approach Message CDN Router DPI Firewall Session Border WAN Controller Accelera7on Carrier Grade NAT Tester/QoE Competitive & Innovative Open Ecosystem Independent Soaware Vendors Orchestrated, automatic & remote install. monitor High volume standard servers SGSN/GGSN PE Router BRAS Radio/Fixed Access Network Nodes • Fragmented non-commodity hardware. • Physical install per appliance per site. • Hardware development large barrier to entry for new vendors, constraining innovation & competition. ETSI NFV High volume standard storage High volume Ethernet switches "Network Functions Virtualisation", NFV ETSI Industry Specification Group, OFC - NFOEC, Mar.17-21, 2013 26 ONF Open Networking Foundation • • Industry organization dedicated to the development, standardization, and promotion of Open Software Defined Networking. Founded in 2011 • • Drive SDN Standards Influence the future of networking Collaborate with the innovators 23 Founding members 90 + membersWho’s w ho in networking and telecommunications Gain visibility in the community Introduction to Open Networking Foundation 2 Education Committee (MEC) Ciena Corporation mcohn@ciena.com Marc Cohn, Chairman Market April 2, 2013 27 ONF ONF Members (March, 2013) • • • • • • • • • • • • • • • • • • • • • • • • 6Wind A10 Networks ADVA Optical Alcatel-Lucent Aricent Argela/Turk Telekom Big Switch Networks Broadcom Brocade Centec Networks Ceragon China Mobile Ciena Cisco Citrix CohesiveFT Colt CompTIA Cyan Optics Dell/Force10 Deutsche Telekom Ericsson ETRI Extreme Networks • • • • • • • • • • • • • • • • • • • • • • F5 Networks/ LineRate Systems Facebook France Telecom Orange Freescale Semi Fujitsu Gigamon Google Goldman Sachs Hitachi HP Huawei IBM Infinera Infoblox Intel Intune Networks IP Infusion Ixia Juniper Networks KDDI Korea Telecom Lancope • • • • • • • • • • • • • • • • • • • • • • • Level3 Comms LSI Luxoft Marvell Mellanox Metaswitch Networks Microsoft Midokura NCL Comms K.K. NEC Netgear Netronome Netscout Systems Nokia Siemens Netw. NoviFlow Oracle Overture Networks NTT Communications Pica8 Plexxi Qosmos Radware Riverbed Technology • • • • • • • • • • • • • • • • • • • • • • Samsung SK Telecom Spirent Sunbay AG Swisscom Tail-f Systems Tekelec Telecom Italia Telefonica Tencent Texas Instruments Thales Tiera Transmode TW Telecom Vello Systems Verisign Verizon VMware/Nicira Xpliant Yahoo! ZTE 4 Introduction to Open Networking Foundation Marc Cohn, Chairman Market Education Committee (MEC) Ciena Corporation mcohn@ciena.com April 2, 2013 28 OpenDayLight http://www.opendaylight.org 29 Deep Programmability within Network • Control-Plane Programmability • • • • Route Control Access Control Network Management Data-Plane Programmability • Packet Data Processing • • • • Cache Transcode DPI Handling New Protocols • • • • Target Scope of OpenFlow/ SDN IPvN (N>6) New Layer2 Irrelevant with OpenFlow (+processors) Content Centric Network (CCN) Meta-Control-Plane Programmability • OpenFlow/SDN with external processors and NFV Defining new proprietary APIs Out of scope of OpenFlow (Included in OpenDayLight?) Scope of Deeply Programmable Network (DPN) Deeper Programmability 30 Deeply Programmable Network(DPN) DPN Programmability for Data Plane NFV Edge Data Processing SDN Programmability for Control Plane via a given API Programmability for defining an API for C/D planes 31 Missing Pieces for Deep Programmability • • • Network Virtualization Software Defined Network (SDN) Network Functions Virtualization (NFV) Further study/research and standardization needed... • • • • Programmability for in-network processing Programmability for new non-IP protocols Programmability for C/D Plane APIs in SDN Accommodation of multiple isolated deeply missing pieces programmable environments 32 Deeply Programmable Network Research in Japan 33 VNode 34 VNode Project in Japan • VNode Infrastructure Enabling Deeply Programmable Network (Project Leader: Aki Nakao) • • 2008-2010 Collaborative Research (NICT/Utokyo/NTT/NEC/Hitachi/Fujitsu) 2011-2014 Collaborative Research (Utokyo/NTT/NEC/Hitachi/Fujitsu/KDDI) The University of Tokyo Confidential 35 R&D on Network Virtualiza;on Technology to Support New-­‐Genera;on Network Develop a technology comprising the service-­‐oriented virtualiza7on pladorm that provides integrated management for various resources including online network, calcula7on and storage and that also can flexibly customize communica7on method, speed, quality and func7ons to establish the new-­‐genera7on network pladorm. Subject C: R&D on network services and applica;ons in NWGN New-­‐Genera;on Sensor Network Applica;on Develop typical network applica7ons that were not easy to build in the past internet environment but would appear in the NWGN era by u7lizing resources in the network. New-­‐Genera;on Contents Distribu;on Applica;on Virtualiza;on Network for Service A Service Programming Cloud Electric Packet Network Virtualiza;on Gateway Data Replica7on Ultra-­‐low Consump;on Power-­‐oriented Network Applica;on Network Applica;on Subject B: R&D on service composable network plaSorm Service Component Management & Implementa;on Network Service Virtualiza;on Integrated Management New-­‐Genera;on Trustable Network Applica;on New-­‐Genera;on Value-­‐added Applica;on Authen7ca 7on Broadca st Develop a service plaSorm that can develop and implement various applica;ons サービスB向け サービスC向け with use of m ixing service parts the network virtualiza;on plaSorm provides. 仮想ネットワーク 仮想ネットワーク Monitori ng Cash Compression Encryp7on Loca7on Info QoS Sta7s7cs Network PlaSorm Network Service Management Virtualized Node Virtualiza;on Virtualized Node Gateway Calculator Lightpath Storage Resource Network Resource Subject A: R&D on integrated-­‐management network virtualiza;on plaSorm Virtualized Node Network Service Management Network Virtualized Node Virtualiza;on Electric Packet PlaSorm Network Virtualiza;on Gateway Server Terminal Virtualiza;on Gateway Lightpath Network Calculator Resource Storage Resource Develop a new-­‐genera7on network virtualiza7on pladorm that can simultaneously provide mul7ple networks being customized by service and that substan7ally enhances its safety, reliability, flexibility etc. to realize high func7onality, large capacity and ultra-­‐low consump7on. (Slide from NICT) 36 VNode Infrastructure (extended to US!) •7 VNodes, 4 Network Connectors, 11 Access Gateways •Deep Programmability for Experimenting with Arbitrary Protocols (Non•Slice-Around-The-World Project (A VNode in U of Utah) AGW Utah Sapporo Kanazawa Sendai Okayama AGW 1G VN Tokyo Area NC International Circuit KR Hiroshima VN AGW AGW 10G 10Gx2 Fukuoka 10G 10Gx2 10G NICT Koganei 40G AGW AGW Tokyo NICT Otemachi NICT Hakusan 40G NC AGW UTokyo International Circuit AGW Nagoya USA TH SG Wireless Testbed Example NC AGW NTT Yokosuka VN AGW VN 40G VN 10G VN 10G AGW 40Gx2 Osaka VN 10G AGW NC VN VNode HK AGW AccessGateway 10G 1G NC Network Connector 37 Data Plane Federation (VNode / GENI) ProtoGENI VNode Okayama Hokuriku Fukuoka Osaka Tokyo1 LA Salt Lake City Packet Cache Slice Nagoya Tokyo2 NC10 LandSat Houston Slice NC7 JGN-X VNode JGN-X VNode Front-End (Network Connector) IDC IDC JGN-X DCN TransPAC3 DCN I2 ProtoGENI IDC Internet2 ION Emulab UEN Internet2 TransPAC3 JP-US circuit Aki Nakao, Plenary Talk @GENI Engineering Conference GEC15, 2012 VNode and GENI Packet Cache Demo File Server KyusyuV VNode Hokuriku Hakusan NC10 PRESTA PRESTA Osaka Data Traffic Hash Traffic Nagoya Otemachi Ingress Cache data sync. Egress Cache pc412 File Download pc350 1.5Mb/s vlan3130 pc551 pc415 300kb/s pc256 pc331 protoGENI University of Utah FLARE 40 FLARE SDN Control Plane Programmability DPN SDN + Data Plane Programmability Deeply Programmable Node Network Virtualiza;on New Protocol Capability 41 FLARE Node Implementation • • • • • • Multiple, isolated, deeply programmable environments Hybrid of many-core processors + x86 processors 1U / 1U Mini Form Factor 4 x 10G, 2x10G + 8x1G (20G non-block switching) Control Plane & Data Plane Linux Programmability Achieve both programmability and performance 42 Ethernet Switch ToGxIO (xgbe2) FromGxIO (xgbe1) Ether Switch ToGxIO (xgbe1) FromGxIO (xgbe2) (Gbps) 10.0 9.9 9.9 9.9 9.9 9.9 9.9 9.9 2 cores 3 cores 4 cores 5 cores 7.8 7.5 5.0 9.9 4.2 2.5 0 1 core Switching Performance pkt_size=512B pkt_size=1514B 43 OpenFlow Switch Control Plane NOX Controller ofprotocol dpctl tunnel Data Plane FromIO (xgbe1) ToIO (xgbe2) OFSwitch FromIO (xgbe2) ToIO (xgbe1) (Gbps) 9.9 10.0 9.9 9.9 7.9 6.7 7.5 2.5 5.1 4.2 5.0 2.7 3.8 1.5 0 1 core 2 cores 3 cores Switching Performance 4 cores 5 cores pkt_size=512B pkt_size=1514B 44 L2 Programmability Extended (96bit) MAC switching ToGxIO (xgbe2) FromGxIO (xgbe1) ExMAC Switch ToGxIO (xgbe1) FromGxIO (xgbe2) (Gbps) 7.5 5.0 9.9 9.7 10.0 9.9 9.9 9.9 9.9 9.9 3 cores 4 cores 5 cores pkt_size=512B pkt_size=1514B 8.7 6.1 3.2 2.5 0 1 core 2 cores Switch Performance Window-based Arbitrary Bit Matching Arbitrary bit matching as in openflow pattern matcher is costly due to expensive memory operation per packet window Set a window to minimize per-packet memory operations Improve performance while keeping flexibility Leon Lee, Ping Du and Akihiro Nakao, "Ouroboros: SDN Beyond Flow-Tuple Matching," IEICE NS Technical Report, Mar. 2013 46 RTP-SSRC Matching Purpose: Specific Real-time audio/video traffic control, based on stream , not flow (e.g. based on RTP SSRC field to enable routing according to the streaming-video IDs ) Benefit: Application/Content specific routing Solution: Use window to extract information included in RTP headers RTP SSRC Payload type …….. 47 Trailer Matching Purpose: Device/application/content specific traffic engineering Benefits: More specific recognition for packets bound to overly used TCP port (e.g., 80) Traffic engineering for data transmitted from specific devices Proposal: After attach/detach trailers, establish control in intermediate FLARE switches CCAV …. 80 HTTP 8080 48 Trailer Matching Trailer is an extra section attached at the end of each packet Headers Packet content Trailer End/Edge nodes are responsible for attaching and detaching Trailers Host1 Host2 Packet Packet FLARE . Packet . Packet . Packet . Leon Lee, Ping Du and Akihiro Nakao, "Ouroboros: SDN Beyond Flow-Tuple Matching," IEICE NS Technical Report, Mar. 2013 49 FLARE at ITPro EXPO 2012 Beyond OpenFlow/SDN 50 MPLS 2012 (with Cisco & Juniper) 11 LivingLab Living with Deeply Programmable FLARE Living#Lab#Control#Plane# 402 Blue#RJ45#ports#(CAT6)# # 7 8 Living&Lab&Data&Plane& F S 9 102 005## L2#SW# F 1 1 005# 402 Fiber&Port& & 2 1 102 J S G F 005& Patch& Panel& F F 2 F FC 105 F 1 107 106 2 F 3 4 5 F 105 6 Blue#RJ45#ports#(CAT6)# # 1 2 1 F F F S 107 106 2 1 2 1 2 Fiber&Port& & 52 Conclusion • Deep Programmability refers to the extensive programmability including Control-plane, Data-plane (including non-IP handling), (re)defining APIs in SDN, etc. • Network virtualization is key to accommodating multiple slices with security-, performance-, namespace-isolated programmability • Standardization for Deeply Programmable Network (DPN) should collectively concepts of Network Virtualization, SDN, NFV, OpenDayLight/OpenFlow, etc is crucial for future network • Sustainable communities can be realized through developing flexible and adaptive Deeply Programmable future network infrastructure against constantly arising problems 53