The improvement of caching strategies about unstructured P2P system

Abstract

Improving cache replacement strategies is an important way to enhance the performance of network. As the lack of cache replacement strategies about P2P systems, the author proposes the corresponding improvement measures and designs a new cache replacement strategy — MFU (Most Frequently Used). This algorithm aims to achieve high-degree peers to delete popular contents and low-degree ones hold popular contents by setting threshold properly, thus reducing the maximum load of peers. And then, using Gnutella system as network model, the author compares MFU with LRU, LFU and ARP through simulation experiment to prove the advantage of hit rate and load balancing about MFU.