ProtocoL for Epidemic protocols Jin Sun Paul Weber Byung Choi Roger Kieckhafer bkchoimtuedu Michigan Technological University 1 9122007 9122007 2 Epidemic Protocols Reliable Broadcasting ID: 932526
Download Presentation The PPT/PDF document "DIMPLE: DynamIc Membership" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
DIMPLE: DynamIc Membership ProtocoL for Epidemic protocols
Jin Sun, Paul Weber, Byung Choi, Roger Kieckhaferbkchoi@mtu.eduMichigan Technological University
1
9/12/2007
Slide29/12/20072
Slide3Epidemic Protocols: Reliable Broadcasting9/12/2007
3
K. Birman, M. Hayden, O. Ozkasap
, Z. Xiao, M.
Budiu
, Y.
Minsky
, “Bimodal Multicast,”
ACM Transactions on Computer Systems, 17(2), 41-88, May 1999.
Slide4BackgroundTwo fundamental assumptions of epidemic protocols
Random selection of next forwardersRandomly at uniform from the entire membershipNodes know the network size (N)Two approachesCentralizedDistributedEntire membership at each and every nodeDifferent partial membership at different nodes
4
9/12/2007
Slide5Epidemic protocols on P2P systems?P2P systems can be very
large in size, and very dynamic in membershipDifficult to maintain a copy of the entire membership at each nodeInconsistencyOverhead Partial membership at each node?
5
9/12/2007
Slide6ChallengesHow to maintain partial membership at each node chosen
randomly at uniform from the entire membership?How to provide different partial membership at different cycles?How to handle dangling pointers
caused by churn?
6
9/12/2007
Slide7General ApproachExchange part of partial membership with part of another partial membership such that, from the node’s perspective, the partial membership is always a random selection of the entire membership at uniform
79/12/2007
Slide8Shuffle!A well known randomization method in gamblingA practical and easy way of randomization
Is it really random at uniform?What should happen if not perfectly random?Network partitioning!8
9/12/2007
Slide9Resilience of ShuffleProbability of network partitioning is diminishingly small,
practically zero!With reasonable sizes ofPartial membership (O(log(N))Shuffle length log(N) Regardless of frequenciesA. Allavena, A. Demers, and J.E. Hopcroft
. “Correctness of a gossip based membership
protocol” ,
24th
ACM Symposium on
Principles of Distributed Computing (PODC ’05
)
.
9
9/12/2007
Slide10Shuffle PropertiesConvergence
Global randomness, comparable to random networksAverage shortest path lengthClustering coefficientRegardless of frequenciesS. Voulgaris, D. Gavidia, and M. van Steen. Cyclon: Inexpensive membership
management for unstructured p2p overlays. Journal of Network and Systems Management, 13(2):197–217, June 2005.
10
9/12/2007
Slide11Shuffle with Churn?Not addressed in the previous
work, huh?Network partition and churn are different!Measurement study on P2P systems shows: Average stay time under 10 minutes! S. Rhea, D. Geels, T.
Roscoe, and J. Kubiatowicz, “Handling Churn in a
DHT,” USENIX
Technical Conference, 2004
.
11
9/12/2007
Slide12Shuffle with Churn?Found not
effectiveProduces a large portion of dangling pointersWould result in poor quality broadcastingMajor reasons:Time delay to detect dangling pointersTime delay in join procedures
9/12/2007
12
Slide13DIMPLE Improves ShuffleReinforcement At the end of each shuffle
Quick JoinUse a list of visited nodes9/12/200713
Slide14DIMPLE-Shuffle works with churn!
149/12/2007
Slide15DIMPLE-Shuffle works with churn!
159/12/2007
Slide16DIMPLE-Shuffle works with churn!16
9/12/2007
Slide17DIMPLE-Shuffle works with churn!
179/12/2007
Slide18DIMPLE: detects dangling pointers fast
189/12/2007
Slide19In-Degree Distribution
199/12/2007
Slide20Out-Degree Distribution
209/12/2007
Slide21DIMPLE: better quality of in-degrees
219/12/2007
Slide22DIMPLE: better quality of out-degrees
229/12/2007
Slide23ConclusionsContributions:DIMPLE makes shuffle to work with churn
A good practical solution to dynamic membership service (for epidemic protocols)Future work:DIMPLE algorithms are improvableNetwork size (N) estimation utilizing shuffle is next!Self-organizing epidemic protocols
23
9/12/2007