A framework for the understanding, optimization and design of structured peer-to-peer systems

Sammanfattning: Peer-to-Peer (P2P) computing is a recent hot topic in theareas of networking and distributed systems. Work on P2Pcomputing was triggered by a number of ad-hoc systems that madethe concept popular. Later, academic research e orts started toinvestigate P2P computing issues based on scientificprinciples. Some of that research produced a number ofstructured P2P systems that were collectively referred to bythe term"Distributed Hash Tables" (DHTs). However, theresearch occurred in a diversified way leading to theappearance of similar concepts yet lacking a commonperspective.In this thesis we present a number of papers representingour reseach results in in the area of DHT systems as thestate-of-theart P2P systems. The contribution of this thesis isthreefold. (i) We present the principle of distributed k-arysearch (DKS) and argue that it serves as a framework for mostof the recent P2P systems known as DHTs. That is, given the DKSframework, understanding existing DHT systems is done simply byseeing how they are instances of that framework. (ii) We arguethat by perceiving systems as instances of the DKS framework,one can optimize some of them. We illustrate that by applyingthe framework to the Chord system, one of the most establishedDHT systems. (iii) We show how the DKS framework helps in thedesign of P2P algorithms by two examples: (a) A broadcastalgorithm that takes advantage of the distributed k-ary searchtree (b) The DKS(n; k; f) system which is a system designedfrom the beginning on the principles of distributed k-arysearch.