A non-partitioned file assignment strategy for parallel i/o system with minimum disk I/O contention probability

Many solutions have been proposed to tackle the problem of assigning nonpartitioned files in parallel I/O systems. The primary objective of the existing solutions is either to balance the load across disks or to minimize the service time variance at each disk, whereas the dynamic file access characteristic such as the disk I/O contention is ignored. As the disk I/O contention may tum the parallel I/O into sequential one, minimizing the disk I/O contention should be an important objective the file assignment strategy should take into account. This paper presents a new static non-partitioned file assignment algorithm named NP-MinCP. The aim of NP-MinCP is to minimize the disk I/O contention probability, thereby optimizing the mean response time of file requests. The experiment results show that NP-MinCP achieves an optimal performance on mean response time among the schemes for comparison. NP-MinCP can be applied to reorganize the nonpartitioned files stored in parallel I/O systems.