Design and Evaluation in Mobility First MobiliytFirst team

Transcription

1 Design and Evaluation in Mobility First MobiliytFirst team Presented by: Jim Kurose Department of Computer Science University of Massachusetts Amherst MA USA

2 Overview v evaluation: architecture, system, prototype v MF evaluation GNS control overhead workloads prototype demos (GENI: end-end, components) other activities v reflection: broader context

3 Architecture, System, Prototype * architecture high-level design/structuring principles, service/function modularity guides, informs, inspires, constrains system instantiated set of interoperating protocols, mechanisms, platforms conforming to design principles here s what it does (function), you tell me how prototype (realization) Implemented (sub)set of protocols, platforms in particular existing technologies * ack: J. Wroclawksi, D. Clark

4 Architecture, System, Prototype telephony Internet ICN architecture end-end circuit, stateless endpoints, stateful core, QoS, single service datagram, stateful endpoints, best-effort stateless core, multiple services, IP content, naming, stateful core, caching system SS7, ESS, MSC, VLR, HLR, GGSN, TCP, UDP, DNS, BGP, IS-IS, OSPF Ongoing (routing, congestion control, caching, name resolution, search, ) prototype many.. over the years many.. over the years Ramping up

5 Architecture, System, Prototype architecture system Evaluation telephony end-end circuit, stateless endpoints, stateful core, QoS, single service SS7, ESS, MSC, VLR, HLD Blocking networks Kleinrock 64 Internet Cerf,&Kahn 74 datagram, Salzer 81 stateful endpoints, Clark 88, best-effort 13 stateless McCanne core, 98 multiple Kelly services 98 Chiang 08 TCP, UDP, DNS, BGP, IS-IS, LOTS OSPF Performance: Queueing networks, delay calculus, effective bandwidths, TCP, NUM optimization Layering: optimization decomposition Arch. complexity Management complexity ICN content, Principles stateful core, enunciated caching Ongoing ongoing, growing (routing, congestion control, caching, name resolution, search, ) simulation, prototyping, some analysis

6 Architecture, System, Prototype architecture system Evaluation telephony end-end circuit, stateless endpoints, stateful core, QoS, single service SS7, ESS, MSC, VLR, HLD Blocking networks Kleinrock 64 Internet MF Cerf,&Kahn 74 datagram, Salzer 81 stateful FIA Inter-architecture content, stateful endpoints, Clark 88, best-effort 13 comparisons core, caching under stateless McCanne core, 98 discussion multiple Kelly services 98 Chiang 08 TCP, UDP, DNS, Ongoing ongoing, (routing, congestion BGP, IS-IS, LOTS OSPF growing Performance: Queueing networks, delay calculus, effective bandwidths, TCP, NUM optimization Layering: optimization decomposition Arch. complexity Management complexity control, caching, name resolution, search, ) simulation, prototyping, analysis, workload analyzing GNS, control, mobility

7 Overview v architecture, system, prototype v MF evaluation GNS: DMap, Auspice control overhead workloads prototype demos (GENI: end-end, components) other activities v reflection: broader context

8 GNS Infrastructure: DMap v approach: hashing GUID to K AS-level resolvers, and local resolver for locally-created GUIDs v evaluation: performance as a function of K simulation: DIMES AS-level topology, latency Weighted query generation, Zipf destination popularity analysis: jellyfish topology model, latency bound DMap: A Shared Hosting Scheme for Dynamic Identifier to Locator Mappings in the Global Internet, Vu, et al, ICDCS 2012

9 GNS Infrastructure: Auspice v approach: active name server replicas, replica controllers, consistency v evaluation: replica placement (local, spatial), # emulation Planetlab, cluster synthetic de novo workload evaluate different replication approaches read/write latency, serverload e+04 1e+05 Sharma, Tie, Uppal, Venkataramani, A Global Name Service for a Highly Mobile Internet Venkataramani, Sharma, Tie, Uppal, Westbrook, Kurose, Raychaudhuri, Design requirements of a global name service for a mobility-centric, trustworthy internetwork, IEEE COMSNETS 2013 Lookup latency (ms) Optimal Auspice CoDoNS Static-3 Replicate-All Load (lookup+update/second)

10 Auspice: Simultaneous mobility recovery latency Server comes up Client comes up 200ms Reconnect complete msocket + Auspice GNS recover from simultaneous mobility in ~2 RTTs, the best achievable latency. UNIVERSITY OF MASSACHUSETTS AMHERST Department of Computer Science 10

11 Overview v architecture, system, prototype v MF evaluation GNS: DMap, Auspice control overhead workloads prototype demos (GENI: end-end, components) other activities v reflection: broader context

12 Evaluating Control Overhead v Question: overhead when routing on names vs. adding a topological address layer? Router-router name exchange: - Hierarchical names - Used for routing packets - Used for caching at routers Hybrid GUID-Name (HGN) Approach: - Use flat GUIDs for caching - Use topological addresses for routing

13 An Analytic Model Model: Nlevels of hierarchy in the name; prefix at level i having l i sub- level prefixes. Define n top {1,2,, N} which indicates the prefix level below which the naming tree starts being influenced by the network topology

14 Sample Result Number of Entries (logscale) L = 10 Routing Table Size with L = 50 Topology Independent Prefixes L = 100 HGN (name independent) Current BGP Table Size n value top Hierarchy in name reduces the table size only when the name prefixes have some degree of dependence on the physical network topology. A. Baid, T. Vu, D. Raychaudhuri, Comparing Alternative Approaches for Networking of Named Objects in the Future Internet, 2012 IEEE Infocom NOMEN

15 Overview v architecture, system, prototype v MF evaluation GNS: DMap, Auspice control overhead workloads prototype demos (GENI: end-end, components) other activities v reflection: broader context

16 Characterizing mobility among networks not your father s mobility: characterize mobility among networks disrnctly different from physical mobility, models physically mobile users may be staronary (from network transiron POV); staronary users may move among networks (mulr- homing, mulrple devices) measure mobility among networks via IMAP logs online users periodically push (background login, check) , and/or intenronally read mail 70 users resident in 183 unique AS numbers

17 Where do users (in aggregate) spend Rme? Work: 5 college Mobile: Verizon, AT&T, sprint) Home: Comcast, Verizon, Verizon, Hughes Misc: 172 other networks in trace Users spend most of their Rme in small number of networks

18 MulR- homing Q: How ofen are users mulr- homed? 15- min subinterval has IMAP access from two different AS s S. Yang, S. Heimlicher, J. Kurose, A. Venkataramani, User TransiRoning Among Networks - a Measurement and Modeling Study, submijed to 2014 IEEE Infocom.

19 Workload, model characterizaron: ongoing mobility measurement on Android devices hjps://nomadlog.net/ generarve network- transironing model: parsimonious MC model for individual user transironing among networks topology models: embedding topology in physical space model exisrng, evolving network topology - Intra- and Inter- AS geo- coverage and connecrons - E.g., test locality- aware protocols, evaluaron future Internet protocol designs capture different external policy effects on network topology evolvement (country connecrvity, gateways) generate syntherc topology models matching measured geo starsrcs

20 Overview architecture, system, prototype MF evaluaron GNS: DMap, Auspice control overhead workloads prototype demos (GENI: end- end, components) other acrvires reflecron: broader context

21 MF Host Protocol Stack Socket API open send send_to recv recv_from close App- 1 App- 2 Network API App- 3 Context API Context Services Linux PC/laptop with WiMAX & WiFi Android device with WiMAX & WiFi Device: HTC Evo 4G, Android v2.3 E2E Transport Network Layer Click Openflow Flare SDN GUID Services Security RouRng Sensors User policies Interface Manager Hop Link Transport Early Dev. Integrate WiFi WiMAX

22 MF MulR- Site GENI Deployment GEC- 16, March 2013 NLR Lincoln, NE Madison, WI Ann Arbor, MI Cambridge, MA Tokyo, Japan Palo Alto, CA Salt Lake, UT N. Brunswick, NJ Los Angeles, CA I2 Atlanta, GA Clemson, SC MobilityFirst RouRng and Name ResoluRon Service Sites MobilityFirst Access Net Long- term (non- GENI) Short- term Wide Area ProtoGENI ProtoGENI

23 Geo- Messaging ApplicaRon with MF Core Services GEC- 18 Oct., 2013 Wisconsin GENI rack Utah GENI rack GENI Internet2 Core BBN GENI rack MF RouRng and Naming Services deployed at 5 GENI rack sites with Internet2 s AL2S providing cross- site layer- 2 connecrvity GENI Edge WiMAX BTS WiMAX BTS MobilityFirst Sofware Router with GNRS instance Dual interface Android phone with WiFi/WiMAX with MF protocol stack ORBIT radio node with WiFi as MF Access point GENI Edge

24 AddiRonal performance- related MF acrvires mulr- homing, mulrpath quanrfying the performance advantages analyzing networks of caches approximate, bounding approaches popularng edge caches: sparal locality

25 Overview architecture, system, prototype MF evaluaron GNS: DMap, Auspice control overhead workloads prototype demos (GENI: end- end, components) other acrvires reflecron: broader context

26 Reflection: broader context architecture system evaluation telephony end-end circuit, stateless endpoints, stateful core, QoS, single service SS7, ESS, MSC, VLR, HLD Blocking networks Kleinrock 64 Internet FIA Cerf,&Kahn 74 datagram, Salzer 81 stateful FIA Inter-architecture content, stateful endpoints, Clark 88, best-effort 13 comparisons core, caching under stateless McCanne core, 98 discussion multiple Kelly services 98 Chiang 08 lots ongoing across TCP, UDP, DNS, FIA, Ongoing broader (routing, congestion BGP, IS-IS, LOTS OSPF community Performance: Queueing networks, delay calculus, effective bandwidths, TCP, NUM optimization Layering: optimization decomposition Arch. complexity control, caching, name resolution, search, ) variety of goals

27 Backup v architecture, system, prototype v MF evaluation GNS: DMap, Auspice control overhead workloads prototype demos (GENI: end-end, components) other activities v reflection: broader context

28 EvaluaKng lookup and update latency UNIVERSITY OF MASSACHUSETTS AMHERST Department of Computer Science 28

29 EvaluaKng load/capacity UNIVERSITY OF MASSACHUSETTS AMHERST Department of Computer Science 29