EFFICIENT CONTENT LOCATION ALGORITHM FOR CONTENT DISTRIBUTION NETWORKS BASED ON DISTRIBUTED CONSTRUCTION OF SEARCH TREE FROM CONTENTS OF PROXIMAL NODES

摘要

Content distribution networks (CDNs) are a type of distributed database using geographically dispersed servers to efficiently distribute large multimedia contents. Among various kinds of CDNs are those resembling peer-to-peer (P2P) networks in which all the servers are equivalent and autonomous, are easy to maintain and tolerant of faults. However, they differ from P2P networks in that the number of nodes joining and leaving the network is negligible. The main problems in CDNs are the placement of contents, and the location of content. Widely used CDNs either have inefficient flooding-like techniques for content location or restrict either content or index placement to use distributed hash tables for efficient content location. However, for the efficient distribution of contents, the contents must be optimally placed within the CDN, and no restrictions should be placed in the content or index placement algorithm. We developed an efficient content location algorithm for CDNs, based on the distributed construction of a search index without imposing any restrictions on the content or index placement algorithm. We described our algorithm, compared it with the existing content location algorithms and showed its effectiveness in increasing the success rate of queries with less traffic.