A Novel Approach for Cooperative Overlay-Maintenance in - - PowerPoint PPT Presentation

▶
a novel approach for cooperative overlay maintenance in
SMART_READER_LITE
LIVE PREVIEW

A Novel Approach for Cooperative Overlay-Maintenance in - - PowerPoint PPT Presentation

Jun 09, 2023 542 likes •905 views

A Novel Approach for Cooperative Overlay-Maintenance in Multi-Overlay Environments 1 Wu-Chun Chung, National Tsing Hua University 2010/11/30 A Novel Approach for Cooperative Overlay-Maintenance in Multi-Overlay Environments Chin-Jung Hsu, CS,

slide-1
SLIDE 1

A Novel Approach for Cooperative Overlay-Maintenance in Multi-Overlay Environments

1

Wu-Chun Chung, National Tsing Hua University 2010/11/30

slide-2
SLIDE 2

Wu-Chun Chung, National Tsing Hua University

2

A Novel Approach for Cooperative Overlay-Maintenance in Multi-Overlay Environments

Chin-Jung Hsu, CS, National Tsing Hua University, Taiwan Wu-Chun Chung, CS, National Tsing Hua University, Taiwan Kuan-Chou Lai, CIS, National Taichung University, Taiwan Kuan-Ching Li, CSIE, Providence University, Taiwan Yeh-Ching Chung, CS, National Tsing Hua University, Taiwan

2010/11/30

slide-3
SLIDE 3

Outline

Introduction Related Work Cooperative Strategy

CFD – failure detection CNPE – network-proximity estimation

Experimental Results Conclusions

Wu-Chun Chung, National Tsing Hua University

3

2010/11/30

slide-4
SLIDE 4

Introduction

2010/11/30 Wu-Chun Chung, National Tsing Hua University

4

 Overlay Network

A virtual network overlay another layer

 Chord, Gnutella, Super-Peer model, etc.

 Focus: over the Internet

140.114.91.88 220.74.26.168 118.169.74.72 209.131.36.158 66.238.93.162 Name IP Port User A 66.238.93.162 80 User B 220.74.26.128 823 User C 118.169.74.72 8080 User D 66.238.93.162 168

. . .

slide-5
SLIDE 5

Internet Visualization

Wu-Chun Chung, National Tsing Hua University

5

2010/11/30

Wikipedia: Opte Project

slide-6
SLIDE 6

Example: Ring-based overlay

Wu-Chun Chung, National Tsing Hua University

6

2010/11/30

Wikipedia: Opte Project

slide-7
SLIDE 7

More and more applications

 Overlay-based applications are growing

 P2P file sharing – gnutella, eDonkey, BitTorrent, etc.  P2P Steaming – PPStream, PPLive, Joost, etc.  Resource Discovery – Mercury, MAAN, etc.  Cloud computing – Cassandra, Hadoop, etc.

 Multiple overlays co-habit the Internet

Wu-Chun Chung, National Tsing Hua University

7

2010/11/30

slide-8
SLIDE 8

A multi-overlay environment (MOE)

2010/11/30 Wu-Chun Chung, National Tsing Hua University

8

slide-9
SLIDE 9

Each overlay network may serve an application

2010/11/30 Wu-Chun Chung, National Tsing Hua University

9

140.114.91.88 220.74.26.168 118.169.74.72 209.131.36.158 66.238.93.162

slide-10
SLIDE 10

Motivation

 Overlay network introduces maintenance cost

 failure detection  latency/bandwidth measurement  routing table adjustment  adaptive approach  … etc.

 n * Cost = large  how to reduce?

Some of these overlay-maintenance costs are

redundant

Wu-Chun Chung, National Tsing Hua University

10

2010/11/30

slide-11
SLIDE 11

Related Work

 [2007][ICDCS] Build One, Get One Free: Leveraging

the Coexistence of Multiple P2P Overlay Networks

 Sharing information to reduce maintenance cost  Focus on two specific overlays

 [2009][DAIS] Exploiting Synergies between

Coexisting Overlays

 A comprehensive consideration on the reduction

  • f maintenance costs

 Lack of the consideration of intersection ratio

2010/11/30 Wu-Chun Chung, National Tsing Hua University

11

slide-12
SLIDE 12

Intersection Ratio

2010/11/30 Wu-Chun Chung, National Tsing Hua University

12

25% Intersection Ratio

Overlay A Overlay B

  • the percentage of nodes which locates in

both overlays

slide-13
SLIDE 13

Objective

  • Multi-overlay environments
  • Reduce the total maintenance cost
  • Propose a general approach
  • Consider a realistic MOE environment

2010/11/30 Wu-Chun Chung, National Tsing Hua University

13

slide-14
SLIDE 14

Cooperative Strategy

2010/11/30 Wu-Chun Chung, National Tsing Hua University

14

n1 n2 n3

Overlay A Overlay B Overlay C

n < n1+n2+n3

Cooperative Maintenance

  • To reduce the maintenance costs
  • The total cost could be smaller
slide-15
SLIDE 15

Master-Slave Model

2010/11/30 Wu-Chun Chung, National Tsing Hua University

15

n1 n2 n3

Overlay A Overlay B Overlay C

n < n1+n2+n3

Cooperative Maintenance Slave Slave Master

  • One overlay is selected to be the master
  • The master overlay could help reduce the common

maintenance operations

slide-16
SLIDE 16

Master-Slave Model

2010/11/30 Wu-Chun Chung, National Tsing Hua University

16

Periodical Maintenance State Sharing

  • Two kinds of inter-overlay protocols to

support two types of overlay maintenance

slide-17
SLIDE 17

Inter-Overlay Protocols

  • Subscribe/Notify protocol

  periodical maintenance  E.g. failure detection

 periodically checks the status of neighbor nodes to ensure the routing mechanism

  • Query/Response protocol

  state sharing  E.g. network-proximity estimation

 share the information of network state to make the decision of routing path

2010/11/30 Wu-Chun Chung, National Tsing Hua University

17

slide-18
SLIDE 18

Cooperative Failure Detection (CFD)

2010/11/30 Wu-Chun Chung, National Tsing Hua University

18

A B

probe

A B

probe

Elimination master slave

slide-19
SLIDE 19

Cooperative Failure Detection (CFD)

2010/11/30 Wu-Chun Chung, National Tsing Hua University

19

A B

probe

A B

probe

Cooperation

C

master slave

slide-20
SLIDE 20

CFD – Subscription Process

2010/11/30 Wu-Chun Chung, National Tsing Hua University

20

slave

A E D B C

probe probe

master

  • 1. subscribe
  • 2. notify

A E D B C

inform forward

slide-21
SLIDE 21

CFD – Notification Process

2010/11/30 Wu-Chun Chung, National Tsing Hua University

21

slave

A E D B C

  • 1. subscribe
  • 2. notify

master

A E D B C

notify probe

slide-22
SLIDE 22

Cooperative Network-Proximity Estimation (CNPE)

2010/11/30 Wu-Chun Chung, National Tsing Hua University

22

A B A ?

Elimination

D C ? ?

master slave

slide-23
SLIDE 23

Cooperative Network-Proximity Estimation (CNPE)

2010/11/30 Wu-Chun Chung, National Tsing Hua University

23

A B A ?

Exploration

D C ? ? E F ? ?

master slave

slide-24
SLIDE 24

CNPE – Query/Response Process I

2010/11/30 Wu-Chun Chung, National Tsing Hua University

24

master

  • 1. query
  • 2. response

A E D B C

slave

A D B C

latency: <50ms(2) 30ms 45ms 20ms 35ms E (30ms), C (45ms)

E

slide-25
SLIDE 25

CNPE – Query/Response Process II

2010/11/30 Wu-Chun Chung, National Tsing Hua University

25

master

  • 1. query
  • 2. response

A E D B C

slave

A D B C

latency: <50ms(3) 30ms 45ms 20ms 35ms E (30ms), C (45ms), B (50ms) Neighbor’s Neighbor

E

slide-26
SLIDE 26

Experimental Environment

  • PeerSim simulator
  • Cycle-based simulation engine
  • Unstructured, Ring, Tree Overlays
  • Parameter K: neighbor numbers
  • Comparison metric: reduction rate

2010/11/30 Wu-Chun Chung, National Tsing Hua University

26

slide-27
SLIDE 27

Comparison Metric –Reduction Ratio

2010/11/30 Wu-Chun Chung, National Tsing Hua University

27

CFD CNPE

  • The higher the reduction ratio is, the more

efficient our approach will be

slide-28
SLIDE 28

Experimental Results –Session Time

2010/11/30 Wu-Chun Chung, National Tsing Hua University

28

0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 20 40 60 80 100 120

Reduction rate Median session time (min.)

Master: Unstructured(K=4), Slave: Ring(K=2) Master: Unstructured(K=4), Slave: Tree(K=3) Master: Ring(K=2), Slave: Unstructured(K=4) Master: Ring(K=2), Slave: Tree(K=3) Master: Tree(K=3), Slave: Unstructured(K=4) Master: Tree(K=3), Slave: Ring(K=2) 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 35.0% 40.0% 45.0% 50.0% 0.0 20.0 40.0 60.0 80.0 100.0 120.0

Reduction rate Median session time (min.)

Master: Proximity(K=4, N=2), Slave: Proximity(K=6, N=2) Master: Proximity(K=6, N=2), Slave: Proximity(K=4, N=2)

CFD CNPE

slide-29
SLIDE 29

Experimental Results –Intersection Ratio

2010/11/30 Wu-Chun Chung, National Tsing Hua University

29

  • 10%

0% 10% 20% 30% 40% 50% 0% 25% 50% 75% 100%

Reduction rate Intersection ratio

Master: Unstructured(K=4), Slave: Unstructured(K=6) Master: Unstructured(K=6), Slave: Unstructured(K=4) 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 0% 25% 50% 75% 100%

Reduction rate Intersection ratio

Master: Proximity(K=4, N=2), Slave: Proximity(K=6, N=2) Master: Proximity(K=6, N=2), Slave: Proximity(K=4, N=2)

CFD CNPE

slide-30
SLIDE 30

CFD + CNPE

2010/11/30 Wu-Chun Chung, National Tsing Hua University

30

Overlay A Unstructured (K=4) Overlay B Unstructured (K=6) Overlay C Ring (K=2) Overlay D Tree (K=3)

50% 25% 75% Master CNPE Master CFD

slide-31
SLIDE 31

CFD + CNPE

2010/11/30 Wu-Chun Chung, National Tsing Hua University

31

24.88% 14.54% 39.42%

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 CFD CNPE Hybrid

Reduction ratio

  • The total reduction rate approximates 40%
slide-32
SLIDE 32

Conclusions

  • Multi-overlay environments have emerged
  • Total maintenance cost is high
  • Some operations are redundant
  • Cooperative maintenance approach
  • A general Master-Slave model

1)

CFD – Subscribe/Notify protocol

2)

CNPE – Query/Response protocol

  • Reduce more than 60%

2010/11/30 Wu-Chun Chung, National Tsing Hua University

32

slide-33
SLIDE 33

Conclusions

Maintain one, Get many free

2010/11/30 Wu-Chun Chung, National Tsing Hua University

33

slide-34
SLIDE 34

Future Work

  • Other operations of overlay maintenance
  • Master overlay selection criteria
  • Automatic selection mechanism

2010/11/30 Wu-Chun Chung, National Tsing Hua University

34

slide-35
SLIDE 35

A Novel Approach for Cooperative Overlay-Maintenance in Multi- Overlay Environments

35

Wu-Chun Chung, National Tsing Hua University 2010/11/30