Capability-Aware Peer Search

Project Summary

We are exploring several research avenues to incorporate richer knowledge about the underlying peer network to enhance peer search. These many techniques are loosely affiliated by their reliance on network-capability awareness, and are all examples of what we call Capability-Aware Peer Search. By leveraging the richer semantics about the underlying network, we hope to provide a balance between the naive flooding-based Gnutella-style search and the more restrictive overlay topologies inherent in structured P2P networks.

Motivation

The peer-to-peer (P2P) model for distributed computing has garnered significant attention in the past few years due to the widespread popularity of file-sharing networks like Morpheus and Gnutella. These systems are noted for their vast size, their self-organizing nature, and the inherent load-balancing and fault-tolerant system behavior they afford.

The research community has focused mainly on two types of P2P networks: structured and unstructured. In structured P2P networks (like Chord, Pastry, Tapestry, and CAN), there are tighter controls over the data placement and topology within the P2P overlay network. One unique characteristics of these tightly controlled P2P systems is that they guarantee location of content, if it exists, within a bounded number of hops. In contrast, unstructured P2P networks (like Gnutella and Freenet) are more loosely controlled and tend to provide a wider range of search capabilities such as search key (search object) identifier or keyword based search; whereas the tightly controlled P2P systems so far only support search by the identifier of the search key. It is well-known that search in current loosely controlled P2P systems, such as Breadth-first search (BFS) used in Gnutella or depth-first search (DFS) used in Freenet is not efficient.

Our goal is to develop capability-aware peer search algorithms that provide the best of both models.

People

Prof. Ling Liu (lingliu@cc)
Weiling Zhu (wzhu@cc)
Mudhakar Srivatsa (mudhakar@cc)
James Caverlee (caverlee@cc)

This research is partially supported by NSF CNS, NSF CCR, NSF ITR, DoE SciDAC, DARPA, CERCS Research Grant, IBM Faculty Award, IBM SUR grant, HP Equipment Grant, and LLNL LDRD.

Any opinions, findings, and conclusions or recommendations expressed in the project material are those of the authors and do not necessarily reflect the views of the sponsors.