Dan Rubenstein's Flash Crowd Research

My work on flash-crowds can be further divided into three areas: P2P support, server collectives, and prediction methods:

• PROOFS: PROOFS is a prototype P2P system I took part in developing. [SRS02] (with an extended version described in [SRS04]) describes the design of this system and evaluates its performance, and includes experiments with our prototype on PlanetLab. There was a cute "mixing" approach to create random graphs with a fairly straightforward undirected search. We also performed a mathematical analysis that demonstrated the scalability of this idea in systems with millions of users in [RS05].
  
  
  • [SRS02] A. Stavrou, D. Rubenstein, and S. Sahu. A Lightweight, Robust P2P System to Handle Flash Crowds. In Proceedings of ICNP'02, Paris, France, November 2002.
    
    
  • [SRS04] Angelos Stavrou, Dan Rubenstein, and Sambit Sahu. A lightweight, robust, p2p system to handle flash crowds. IEEE Journal on Selected Areas in Communications (JSAC), special issue on Service Overlay Networks, 22(1), January 2004.
    
    
  • [RS05] Dan Rubenstein and Sambit Sahu. Can unstructured p2p protocols survive flash crowds? IEEE/ACM Transactions on Networking, 13(3), June 2005.
  
  
  
• Server Collectives: in [VR03], we analyzed when it made sense for independent service providers to form collectives, pooling resources to host one another's content. Our queueing analyses showed that as long as the intensities (ratio of demand to service capacity) of the providers was similar, they could all benefit from sharing, but slight deviations would make the collective unappealing to the more heavily-loaded providers. We showed that setting tresholds (limiting how much of a provider's resources would be shared) increased this deviation, but it still remained small. The work was extended in [VR05].
  
  
  
  • [VR03] Daniel Villela and Dan Rubenstein. A queueing analysis of server sharing collectives for content distribution. In Proceedings of the Eleventh IEEE International Workshop on Quality of Service (IWQoS 2003), June 2003.
    
    
  • [VR05] Daniel Villela and Dan Rubenstein, Performance Analysis of Server Sharing Collectives for Content Distribution, IEEE Transactions on Parallel and Distributed Systems, 16(12),December, 2005.
  
  
  
• Flash Crowd Prediction: we explored the feasibility of prediction methods for flash crowds. We show in [BCPRSY05] that simple predictive methods, such as linear-fit extrapoloation, will often, far in advance, correctly identify the impending arrival of a flash crowd.
  
  
  
  • [BCPRSY05] Yuliy Baryshnikov, Ed G. Coffman, Guillaume Pierre, Dan Rubenstein, Mark Squillante, and Teddy Yimwadsana. Predictability of web-server traffic congestion. In 10th International Workshop on Web Content Caching and Distribution (WCW 2005), Sophia Antipolis, France, September 2005.