Presentation is loading. Please wait.

Presentation is loading. Please wait.

Community Clustering in Distributed Publish/Subscribe System Wei Li 1,2,Songlin Hu 1, Jintao Li 1, Hans-Arno Jacobsen 3 1 Institute of Computing Technology,

Similar presentations


Presentation on theme: "Community Clustering in Distributed Publish/Subscribe System Wei Li 1,2,Songlin Hu 1, Jintao Li 1, Hans-Arno Jacobsen 3 1 Institute of Computing Technology,"— Presentation transcript:

1 Community Clustering in Distributed Publish/Subscribe System Wei Li 1,2,Songlin Hu 1, Jintao Li 1, Hans-Arno Jacobsen 3 1 Institute of Computing Technology, Chinese Academy of Sciences 2 Graduate University of Chinese Academy of Sciences, Beijing, China 3 University of Toronto, Toronto, Canada IEEE Cluster 2012

2 Agenda Background Algorithms Experiments Conclusions

3 Background Distributed publish/subscribe systems  Clients (publishers & subscribers)  Routers (a.k.a. brokers) … Distributed Router System … Advertisement

4 Background … Distributed Router System … Subscription Advertisement

5 Background … Distributed Router System … Advertisement Subscription Publication

6 Background Distributed Publish/Subscribe Systems  Loosely coupled communication abstraction  Widely used in industry, for example GooPS at Google PNUTS at Yahoo!

7 Client Placement Client placement affects performance of the system  Current solutions Connecting to closest broker [Chen_05] Interest clustering of subscribers [Querzoni_08, Riabov_02] Publisher dynamic placement [Cheung_10]  Limitations Complex communication relationships in interacting clients are not considered The cost of client relocation is not considered

8 Algorithms Problem definition  Network of interacting clients  Distributed routers

9 Algorithms Problem definition cont’d.  The allocation of clients to routers Maximize the performance of the system Minimize the cost of client allocation

10 Agenda Background Algorithms Experiments Conclusions

11 Algorithms Overview

12 Algorithms Steps  Phase 1: Network construction among clients  Phase 2: Community division of client network Newman’s algorithm: modularity-based [Newman_04]

13 Algorithms Steps  Phase 3: Heuristic community clustering Majority-place Mp:

14 Algorithms Steps  Phase 2 and Phase 3 are iterative: Re-divide several communities into smaller ones Performance lose vs. deployment cost decrease Achieve trade off between performance and deployment cost  Phase 4: Load balancing

15 Agenda Background Algorithms Experiments Conclusions

16 Experiments Community clustering vs. interest clustering Experiment settings  Different relationship modes of clients Random Small-world Scale-free  Differently structured router overlays

17 Evaluation Different relationship modes among clients  Message distribution

18 Evaluation Different relationship modes among clients  Message latency & load reduction

19 Evaluation Different cluster compositions

20 Agenda Background Algorithms Experiments Conclusions

21 A community clustering method is proposed for distributed publish/subscribe systems Community clustering is effective to improve the performance under different experimental settings


Download ppt "Community Clustering in Distributed Publish/Subscribe System Wei Li 1,2,Songlin Hu 1, Jintao Li 1, Hans-Arno Jacobsen 3 1 Institute of Computing Technology,"

Similar presentations


Ads by Google