Paper #: 5.1.1
Title: Analysis of the Evolution of Peer-to-Peer Systems

(1) Problems

This paper addresses the maintenance protocol required of any realistic peer-to-peer system. As new nodes join in and others leave, the system must maintain the integrity of the network in order to survive. Several current routing protocols have unrealistic models of maintenance, and this paper presents a method of real-time updating that falls within boundaries of bandwidth that are efficient, while guaranteeing the integrity of the system (given certain assumptions, like a predictable maximum fail rate and join rate).  In particular, the Chord network is examined, with a few proposed modifications.

(2) New Idea and Strengths

Being unfamiliar with peer-to-peer networks and the language used to describe them, I am making a few assumptions here. It seems that this paper is introducing a few terms to define the state of the Chord network, in order to prove that this new idealization protocol will return the network to an ideal state (which they define). The protocol is proven to work on an order of log N, although they point out the system will probably only approach the ideal state, as any additions / subtractions will make the network non-ideal. They do prove that assuming a fairly safe rate of joins and a reasonable number of failures, the iterative idealization process will eventually stabilize the network and continually bring it closer to the ideal state.

(3) Weaknesses and Extensions

It is a little unclear from this paper where the current Chord system ends and their improvements begin. The mathematical proofs for their lower-bound half-life declarations are a bit thick for me, and I am assuming they are correct. The notion of periodically checking for the nodesĄ¯ predecessors and successors in order to maintain a quickly searchable ordering of keys doesnĄ¯t seem particularly revolutionary to me. This paper felt like they were mainly trying to prove that this method would successfully work under the timelines initially set forth, rather than finding an ideal method of updating.