Regenerating Code Based P2P Storage Scheme with Caching

Abstract

High data availability is an important feature of P2P storage system. Erasure coding is a common method employed to support the property. But the technique requires an original file to produce redundant data. On the contrary Regenerating Code, without maintaining the original file, solves the problem through collecting encoded information in a distributed way. However, there is a drawback in this method; that is, encoding needs more peers to get involved in order to obtain good information, which is a harsh condition in a P2P environment due to the requirement of keeping more peers alive simultaneously to hold the file blocks. On the condition that the system did not need to store a replica, we recorded the information of peers which have accessed files recently and utilized the data which have been accessed last in the LRU cache of the peer to improve the access performance and reduce the encoding cost. Through simulation, we experimented with different cache sizes under various peer availability. The results showed that our scheme can successfully improve at least 83 percent of the access performance and lower the cost of Regenerating Code when a file is divided into 7 blocks and the cache size is 64 file blocks.