PEER : P eer-to-peer E nhanced E dge R outer Patrick Crowley This - - PowerPoint PPT Presentation

▶
peer p eer to peer e nhanced e dge r outer
SMART_READER_LITE
LIVE PREVIEW

PEER : P eer-to-peer E nhanced E dge R outer Patrick Crowley This - - PowerPoint PPT Presentation

Nov 16, 2022 483 likes •662 views

PEER : P eer-to-peer E nhanced E dge R outer Patrick Crowley This is the work of my doctoral student Shakir James 2009 Intel Embedded and Communications Education Summit P2P: Two Points of View A users point of view Support many

slide-1
SLIDE 1

PEER: Peer-to-peer Enhanced Edge Router

Patrick Crowley

This is the work of my doctoral student Shakir James

2009 Intel Embedded and Communications Education Summit

slide-2
SLIDE 2

2 - Patrick Crowley – Feb 2009

P2P: Two Points of View

 A user’s point of view

  • Support many applications
  • Offer “inexpensive” scalability
  • Recover quickly from failures

 An ISP’s point of view

  • Route traffic over costly transit links
  • Increase broadband customers, but
  • Surge in traffic ≠ surge in $$$
  • Image from talk "P2P: An ISP’s Point
  • f View," by Pablo Rodriguez
slide-3
SLIDE 3

3 - Patrick Crowley – Feb 2009

The Problem

 Duality of P2P

  • Cheap for content providers, but
  • Expensive for ISPs

 “Cat and mouse” game

  • 1. ISPs: Install traffic-shaping devices
  • 2. P2P : Obfuscate traffic
  • 3. Repeat until…

 No end in sight!

  • FCC forced ISPs to (temporarily) capitulate
  • Damaged relationship in long-term
slide-4
SLIDE 4

4 - Patrick Crowley – Feb 2009

Our Goals

 Build a network device that

  • Controls costs for the ISP, and
  • Maintains good performance for end-users

 Show that ISPs can take unilateral action to

  • Foster a sustainable co-existence with P2P
  • Take the first step in fixing relationship

 Two issues to resolve

  • Illegal content? DMCA “Safe Harbor” a la YouTube
  • How does P2P increase costs for ISPs?
slide-5
SLIDE 5

5 - Patrick Crowley – Feb 2009

ISP Economics

Access Transit Internet

ISP ISP ISP ISP ISP

 Client-server economics

  • $$
  • $$
  • $$
  • $$

server client transit link

slide-6
SLIDE 6

6 - Patrick Crowley – Feb 2009

Access Transit Internet

ISP ISP ISP ISP ISP

ISP Economics

 P2P economics

  • $$
  • $$
  • $$
  • $$

peer peer transit link ISP-local peer Bob

slide-7
SLIDE 7

7 - Patrick Crowley – Feb 2009

BitTorrent Operation

  • Swarm
  • GradsGoneWild.xvid
  • GradsGoneWild.torrent

web server tracker new peer leecher seeder

slide-8
SLIDE 8

8 - Patrick Crowley – Feb 2009

PEER - 1

 P2P Enhanced Edge Router

  • Reduces costly, transit traffic
  • Maintains client performance

 ISPs install PEER alongside edge routers to

  • Exploit existing locality within its network
  • Reply to tracker queries with only local peers

 PEER works on control traffic by

  • Detect peer-to-tracker query messages
  • Extract metadata from messages
  • Respond to peer with a list of only local peers
slide-9
SLIDE 9

9 - Patrick Crowley – Feb 2009

Access Transit

ISP ISP ISP

PEER - 2

$$ $$ PEER installed

 First-reply policy

Bob

slide-10
SLIDE 10

10 - Patrick Crowley – Feb 2009

PEER - 3

  • Parser
  • Oracle
  • Proxy

Cache

  • Filter

incoming packets detect control messages forward message extract metadata reply or forward ? construct response send response

  • .
  • .
  • .
slide-11
SLIDE 11

11 - Patrick Crowley – Feb 2009

Evaluation -1

 Prototype implementation

  • Use programmable routers
  • Deploy on the Open Network Lab (ONL) as a plugin

 Goal

  • Compare transit traffic with PEER and without

 Metrics

  • Upstream/downstream utilization
  • User download time

 Flash-crowd workload

slide-12
SLIDE 12

12 - Patrick Crowley – Feb 2009

Evaluation -2

Factors Level -1 Level 1 Tracker, choice of Application PEER Number of leechers 20 200 Degree of locality Low High Peer behavior Disconnect immediately Stay connected File size 10 MB 100 MB 1st seeder location External Local

 Factors and their levels in our experimental

design

 26-2 fractional factorial design

  • 16 instead of 64 experiments – saves time!
slide-13
SLIDE 13

13 - Patrick Crowley – Feb 2009

Evaluation -3

Effect % variation (var) explained Upstream Downstream Download time Tracker 7.91 8.39 0.00

  • No. of leechers

59.99 32.92 3.05

  • Deg. of locality

5.28 0.31 0.00 Peer behavior 1.98 1.52 0.83 File size 1.24 9.88 94.97 1st seed location 0.23 11.62 0.00 Tracker & locality 7.36 1.12 0.03

 Download time: 95% of var in is due to file

size, and the tracker is responsible for 0%.

 No. of leechers account for majority of

upstream and downstream utilization.

 For upstream, the tracker, degree of locality

and the interaction between them are important.

 For downstream, the initial location of the

seeder and interaction between the tracker and no. of leechers (10.03, not shown).

slide-14
SLIDE 14

14 - Patrick Crowley – Feb 2009

Evaluation -4

perfect localization

slide-15
SLIDE 15

15 - Patrick Crowley – Feb 2009

Related Work

 ISP-dependent: controlled by ISPs

  • Traffic-shaping devices
  • Caching content

 Developer-dependent

  • Biased-neighbor selection
  • Network coordinates

 Co-operative

  • ISP-managed “tracker” service
slide-16
SLIDE 16

16 - Patrick Crowley – Feb 2009

Concluding Remarks

 PEER is a network device that

  • Improves P2P network utilization, and
  • Maintains client performance

 It works across the board and out of the box!  Next steps…

  • Explore other response policies
  • Deploy PEER on university network to address recent

complaint about excess P2P usage, albeit all legal.

  • Implementations on Atom, Tolopai
slide-17
SLIDE 17

17 - Patrick Crowley – Feb 2009

Thanks for listening!

 Questions?