Less Pain, Most of the Gain: Incrementally Deployable ICN

Transcription

1 Less Pain, Most of the Gain: Incrementally Deployable ICN Seyed K. Fayazbakhsh, Yin Lin, Amin Tootoonchian, Ali Ghodsi, Teemu Koponen, Bruce Maggs, KC Ng, Vyas Sekar, ScoE Shenker 1

2 A high- level view of ICN S1 e.g., CCN, DONA, NDN, 4WARD. C S2 C C Decouple what from where Bind content names to intent Today: Fetch from server IP Equip network with content caches Route based on content names e.g., find nearest replica 2

3 Gains of deploying ICN e.g., CCN, DONA, NDN, 4WARD. C C Lower latency Reduced congeszon Support for mobility Intrinsic security 3

4 Pains of deploying ICN NDN, 4WARD. e.g., CCN, DONA, C C Routers need to be upgraded RouZng needs to be content based 4

5 MoZvaZon for this work Gains Lower latency Reduced congeszon Support for mobility Intrinsic security Can we get ICN gains without the pains? e.g., exiszng technologies? e.g., incrementally deployable? Pains Routers need to be upgraded with caches RouZng needs to be content based 5

6 Approach: AEribute gains to tenets QuanZtaZve QualitaZve Lower latency Reduced congeszon Support for mobility Intrinsic security Decouple what from where Bind content names to intent Equip network with content caches Route based on content names 6

7 Key Takeaways To achieve quanztazve benefits: à Just cache at the edge à With Zipf- like workloads, pervasive caching and nearest- replica rouzng don t add much To achieve qualitazve benefits: à Build on HTTP Basis for incrementally deployable ICN 7

8 Outline Background and Approach Analyzing quan,ta,ve benefits QualitaZve benefits à Incrementally deployable ICN Discussion 8

9 Design space of caching QuanZaZve benefits are largely due to caching Two key dimensions to this design space: Cache placement E.g., everywhere? Edge? Request rouzng E.g., shortest path, nearest replica? 9

10 RepresentaZve points in design space Cache Placement Request RouZng ICN- SP Everywhere Shortest path to origin ICN- NR Everywhere Nearest replica Edge Only at edge nodes Shortest path to origin Edge- Coop Only at edge nodes Shortest path to origin Edge neighbors alone 10

11 SimulaZon setup Edge Real CDN request logs Cache provisioning ~ 5% of objects Uniform or ProporZonal LRU replacement PoP- level topologies (Rockecuel) augmented with access trees Assume name- based rouzng, lookup incurs zero cost 11

12 % improvement over no- cache Query latency improvement (%) Request latency ICN-SP ICN-NR EDGE EDGE-Coop 0 Telstra Sprint Level3 AT&T Gap between architectures is small (< 10%) Similar results for congeszon + server load 12

13 SensiZvity Analysis % gap ICN- NR - Edge Performance gain of ICN-NR over EDGE (%) Latency Congestion Origin-Load -5 Baseline Best case Normalize Double Even in best case, ICN- NR is only 17% beeer Gap can be easily reduced 13

14 ImplicaZons of Edge Caching Incrementally deployable Domains get benefits without relying on others IncenZve deployable Domains users get benefits if domain deploys caches 14

15 Background and mozvazon Outline Approach QuanZtaZve benefits of ICN Qualita,ve benefits à Incrementally deployable ICN Discussion 15

16 RevisiZng QualitaZve Aspects 1. Decouple names from locazons Build on HTTP Can be viewed as providing get- by- name abstraczon Can reuse exiszng web protocols (e.g., proxy discovery) 2. Binding names to intents Use self- cerzfying names e.g., Magnet URI schemes Extend HTTP for crypto and other metadata 16

17 idicn: Content RegistraZon Name ResoluZon System Register L.P.idicn.org P = Hash of public key L = content label e.g., hep://en.5671.fda627b.idicn.org/wiki/ Reverse Proxy Publish content Origin Server 17

18 idicn: Client ConfiguraZon Name ResoluZon System Proxy Edge Cache Reverse Proxy AutomaZc Proxy Discovery e.g., WPAD Client Origin Server 18

19 Proxy Edge Cache 2. Name resoluzon idicn: Content Delivery Name ResoluZon System 3. Rqst by address 5. Response + Metadata Try it out: Reverse Proxy 1. Rqst L.P.idicn.org 6. Response 4. Fetch Client Origin Server 19

20 Conclusions MoZvaZon: Gains of ICN with less pain Latency, congeszon, security Without changes to routers or rouzng! End- to- end argument applied to ICN design space Can get most quanztazve benefits with edge soluzons Pervasive caching, nearest- replica rouzng not needed Can get qualitazve benefits with exiszng techniques With exiszng HTTP + HTTP- based extensions Incrementally deployable + backwards compazble idicn design: one possible feasible realizazon Open issues: economics, other benefits, future workloads.. 20