1 / 35

MAP-Me Managing Anchor-less Producer Mobility in Content-Centric Networks

MAP-Me Managing Anchor-less Producer Mobility in Content-Centric Networks. Jordan Augé , Software Engineer, Cisco Systems (France ). joint work with G. Carofiglio , G.Grassi , L. Muscariello , M. Papalini , G.Pau , X. Zeng. ICNRG interim meering , March 18 , 2018.

fulbright
Download Presentation

MAP-Me Managing Anchor-less Producer Mobility in Content-Centric Networks

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. MAP-MeManaging Anchor-less Producer Mobility in Content-Centric Networks Jordan Augé, Software Engineer, Cisco Systems (France) joint work with G. Carofiglio, G.Grassi, L. Muscariello, M. Papalini, G.Pau, X. Zeng ICNRG interim meering, March 18, 2018

  2. Mobility challenges Mobility challenges beyond ICN • Coreoverloadedwithsignalization and user traffic • Opportunity to removeanchors, tunnels… simplify & unify • Much interest in IETF (DMM/ILA/…) and 3GPP Interesting ICN featuresfrom a mobilitystandpoint Much progress in producermobility… … and some original solutions wrt IP

  3. Consumer vs. Producer mobility • Consumer mobilitynativelysupported in virtue of the pull-based model • Producer Mobility = still an open challenge We are onlylooking at L3 here, transport need to support / leverage ICN specifics, and help withseamlesshandovers Use socket library + forwardermechanisms as in: ! G.Carofiglio, L.Muscariello, M.Papalini, N.Rozhnova, X.Zeng, “Leveraging ICN In-network Control for Loss Detection and Recovery in Wireless Mobile networks”, In. Proc. ICN’16, Sep. 26-28, 2016, Kyoto (JP)

  4. Towards native mobility • Preservefeatures of ICN (caching, multipath, content-basedsecurity) : no tunnelling, no name rewriting • L2-agnostic • Decentralized: remove the need for anchors • Feasible and scalable: lightweight in terms of network / CPU usage • High bandwidth / lowlatencyrequirements • Preserve performance of user flows in progress

  5. Illustration of relatedapproaches in ICN P P P P C C C C C P RDV-BASED ANCHOR-BASED TRACE-BASED ANCHOR-LESS [DNS] [SCOM, SNAMP, …] [Mobile IP] [MobiCCN, …] [LBMA, KITE, …] [InterestForwarding, …]

  6. MAP-Me in a nutshell • Use of data plane protocolsfor sendingforwarding updates to the network • Producer sendsmobility updates to itself • Modifies the forwarding graph by patching a small set of routers’ FIB entries (nexthops) • → Update protocolwith good path stretch properties for fast global reconnectivity : routingcanoptimizepathslater… + Notification protocol: Optimizations for low-latencytraffic

  7. MAP-Me implementationoverview 3. Processing 2. Packet formats 4. Temporary state 1. Mobility detection & prefix management

  8. 1. Mobilitydetection & prefix management Mobilitydetection A FaceManager module listens to L2/L3 events and dynamicallycreate/delete faces. Example on Linux: Prefix management Hook the FIB to learn about locallyservedprefixes Netlink Ctrl. Protocol Forwarder FaceManager Linux 802.11 [Create Face] [new AP/IP neigh]

  9. 2. Packet formats sequence number Interest (I.) I. Seq Name I. Update I. Notification Seq Prefix Sec. T. Seq IU Prefix Sec. T. IN I. Update Ack I. Notification Ack Seq Prefix Sec. T. INack Seq Prefix Sec. T. IUack

  10. 3. Processing: Update protocol Modify FIB entries at forwardingtimescale to min. disconnectivity assumption ▪ existence of a routingprotocolpopulatingFIBs ▪ multipath support (not shownhere for clarity) key ideas ▪ interest-basedsignalingtriggered by produceraftermovement▪ reusestale FIB entries to forwardsignalization ▪ flip edgesin existingforwardingtree to forward to new location

  11. Update propagation 0 1 0 0 0 1 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 0 0 0 0 0 IU(1) IU(1) IU(1) IUack(1) IUack(1) IUack(1) 1 0 0 1 1 0 1 1 0 0 0 0 0 Hop-by-hop retransmissions for reliability 0 0 0 0 0 0 0 0 0 0 0 0

  12. Concurrent updates 0 2 1 0 0 0 0 0 0 2 0 0 0 ⚠ 0 0 0 2 2 0 2 0 2 0 0 0 0 0 0 0 0 0 0 0 2 2 0 1 1 2 1 0 0 0 2 0 2 2 2 2 IUack(2) IUack(2) IU(1) IU(1) IU(2) IUack(2) IUack(1) IU(2) IU(2) IU(2) IUack(2) 1 0 0 1 1 0 2 2 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

  13. 3. Processing : Notification protocol 0 0 3 0 2 2 1 1 0 2 0 1

  14. 4. Transient FIB (TFIB) Fj Fi Fp Fm m sequence number nexthop faces IUack(2) IU(2) retxtimers 1 2 /prefix [ Fm ] { Fk→ τm} k IU(2) i IUack(2) 1 2 TFIB j Fk p n

  15. Implementationsummary 3. IU/IN processing Scopeddiscovery • 4. TFIB • Sequencenumber • Reliability (ReTX) 1. Face Manager FIB inspection 2. IU / IN and ack IU IN IN Ack IU Ack

  16. MAP-Me timescales e.g. 5s e.g. 1m FAST MOBILITY ROAMING MOBILE RELOCATION Notifications Updates Routing TR TU improves ? Near-zero disconnectivity Fastforwarding plane restoration Recomputation of optimal paths Update 1 node Update 1 path Update whole graph

  17. Interaction between MAP-Me and routing FIB update overwritten by routingprotocol ? Initial proposal: independentprotocols • a routing instance running on producer • seq → (Rseq, seq) • delay MAP-Me IU/IN until local routing convergence • reissue IU/IN afterrouting update has been received FIB … Routing Protocol (e.g. ISIS) [node/linkfailure] MAP-Me [mobility] Detailedexamplewithlink state in paper Towards a joint protocol ?

  18. MAP-Me properties [seepaper] for UPDATES • Optimal reconnectivityalgorithm; bounded stretch properties • Update Algorithmcorrectness and convergence have been proved • Preserves the loop-freeness of the forwardingtree • Works in presence of multipath for notifications • Near-zerodisconnectivity • Ability to use variousheuristics for instance to bound stretch / load of neighbouring links, etc.

  19. Security • A lightweightprefix attestation schemebased on hash-chains • Appropriate for Trace-based and Anchor-lessmobilityapproaches • Preserves MAP-Me benefits in terms of user / network performance • Lightweight in term of CPU • Additionalsecuritywrttypical HMAC basedapproaches in a distributedcontext A. Compagno, X. Zeng, L. Muscariello, G. Carofiglio, J. Augé, “Secure Producer Mobility in Information Centric Networking”, in. Proc. ICN’17, Sep. 26-28, 2017, Berlin (DE)

  20. Evaluation

  21. Evaluation methodology Analysis Simple analytical model under RWP: MapMe> AB wrt. overheadand offloading Graph simulations Synthetic / Rocketfuel topologies : stretch Simulation Synthetic & Trace-based - user & network performance Emulation Demo & testbed evaluations (Grid5k → Cisco)

  22. ndnSIM simulation setup ▪ Realistic 802.11n radio access(handover, rate adaptation, fading, etc.) ▪ N permanent CBR pairs ▪ Randomwaypointmobility (uniformlydist. and varyingconst. speed) &vehicular trace ▪ Baseline scenario: - C = 10Mb/s – 5ms delay - N = 1..5..20 - speed: up to 50km/h 80m

  23. Simulations

  24. Simulations

  25. Behaviour of mobilityprotocols Alwaysconnected Somepath have longer stretch… Hop update Time to reach Anchor Impact of routing !

  26. Avg. stretch and latency on synthetic topologies

  27. Conclusion MapMe: effective anchorlessproducermobility solution for ICN • improvementsover state-of-the-art solutions • simple, easilydeployable Future workincludes: • WiFi / LTE handover- Interdomain • Multi-source / Multihoming • More experimentation (seenextpresentation) J.Augé, G.Carofiglio, G.Grassi, L.Muscariello, G.Pauand X.Zeng, MAP-Me: Managing Anchor-less Producer Mobility in Content-CentricNetworks, in IEEE TNSM, vol. PP, no. 99.

  28. MAP-Me implementation & deploymentCICN, NDN, Hybrid ICN (hICN) Jordan Augé, Software Engineer, Cisco Systems (France) joint work with G. Carofiglio, L. Muscariello, M. Papalini, M. Sardara ICNRG interim meering, March 18, 2018

  29. NDN prototype • Original C++ / boostimplementationused in paper • Compatible bothwith NFD and NDNSIM • Link to paper and NFD source code : http://mapme-tnsm.github.io/ • Not maintained • Bonus : implementations of alternative solutions: • Anchor-based (generic) • Trace-based (inspiredfrom KITE)

  30. « MakingWiFi mobile » demo @ MWC’17 DEMONSTRATED SCENARII➊ CONSUMER MOBILITY➋ ASYNC. MULTICAST➌ NATIVE MULTIHOMING Server ICN backhaul ➍ PRODUCER MOBILITY & LIVE VIDEO STREAMING ICN routers APs

  31. CICN (hICN) • Continuation of CCNx’sMetisforwarder • C / libparc (data structures: hashmaps, etc.) • Most codebaseidenticalbetween CICN & hICN • Differences at lowerlayers (packet formats, etc) • Veryfresh, willbereleased in fd.io shortly • No notifications yet • No security

  32. MAP-Me in Hybrid ICN Uponattachment to P1, producer sens an IU thatshould update forwarding state in bothhICN and IP routers on the way to P0. hICNprinciples of reusing RFC compliantpackets. In hICN, wecan use an non-reactiveICMP(v6) redirectpacketto P0’s addressusing the link local address as default gateway. Sequencenumbercanbeencoded in payload, and securitycanbeensured as usual by AH header, SEND, etc. C P 0 hICN ROUTER IP ROUTER IU IP ROUTER P 1 hICN ROUTER

  33. ICMP redirectprocessing In an hICN router • ICMP-RD packetsrecognized and used to update FIB entries • Seq. check, multipath possible In an IP router • IPv6 routerscanbeconfigured to use ND proxy thatwill update local route cache (withhigher administrative distance), rewrite the gatewaywithown LL address, and forwardit. • no multipath • no local check on seqno., deferred to closesthICN router [1]

  34. ExperimentationwithvICN J.Augé, G.Carofiglio, G.Grassi, L.Muscariello, G.Pauand X.Zeng, MAP-Me: Managing Anchor-less Producer Mobility in Content-CentricNetworks, in IEEE TNSM, vol. PP, no. 99, pp. 1-1.

  35. Conclusion • Two ICN implementations • ICN (opensource) and hICN • We have justfinished a draft for MAP-Me • working on hICN version • Wewelcomecommunity discussion and feedback • MAP-Me ~ a « mobility control plane » for (Hybrid) ICN • Interestinganchorlessproperties are generally applicable • Interestbeyond the ICN community ? CICN project and CICN source code: http://wiki.fd.io/view/Cicn Link to paper and NFD source code : http://mapme-tnsm.github.io/

More Related