1 / 8

Can Heterogeneity Make Gnutella Scalable?

Can Heterogeneity Make Gnutella Scalable?. Gisik Kwon Dept. of Computer Science and Engineering Arizona State University. Motivation. Scalability of Gnutella TTL-basd flooding Pros: easily accommodate highly transient node population Cons No guarantee to locate existing file

afric
Télécharger la présentation

Can Heterogeneity Make Gnutella Scalable?

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Can Heterogeneity Make Gnutella Scalable? Gisik Kwon Dept. of Computer Science and Engineering Arizona State University

  2. Motivation • Scalability of Gnutella • TTL-basd flooding • Pros: easily accommodate highly transient node population • Cons • No guarantee to locate existing file • No scalablility • Improving scalability • Higher capacity nodes carry the heavier burdens • Random walks searching

  3. Design • Main features • Flow control: Restrict flow of queries into each nodes • Topology adaptation: query flow toward a sufficient capacity node • Terms • Ci: maximum # messages node I is able to process over a given time interval T • In[j,i]: # incoming messages from node j to I in the last time interval T • Out[I,j]: # outgoing messages from node i to j in the last time interval T • outMax[I,j]: maximum # messages node I can send to node j per time interval T • Out[I,j] <= outMax[I,j] <= out[I,j] + (Cj – in[*,j])

  4. Pseudo code

  5. Evaluation • Setup • =10, =1.25, T=100 seconds • Object popularity: Zipf-like distribution( = 1.2) • Query rate: poisson process(1.2 queries/min) • Node heterogeneity • Ci: Zipf-like distribution( = 2.0) • Bandwidth distribution: from measurement • Dial-up modem(a fair fraction), cable or DSL(majority), high speed(small portion) • Uniform random graph topology • For Zipf-like capacity distribution • 10000 nodes, avg. degree 9.9 • For Gnutella-like capacity distribution • 5000 nodes, avg. degree 7.5

  6. Average query resolution time

  7. Query resolution

  8. Degree distribution

More Related