A Novel Cluster Parallel File System

We propose a novel cluster parallel file system: high-throughput & scalable parallel file system (HTSPFS). There are some new features in it such as automatic adaptive file striping policy based on file size, request arrival rate, concurrent of accessing files, storage servers' networking and disk performance; prefetching read-request and writing behind write-request to enable a higher throughput in HTSPFS. We give a detail discuss of Markov spatial predicting method for read request, experiment results show that Markov predicting technology has high prediction veracity. Overall, HTSPFS has some autonomies characteristics