Mapping Functions and Data Redistribution for Parallel Files

A parallel file may be physically stored on several independent disks and logically partitioned by several processors. This paper presents general algorithms for mapping between two arbitrary distributions of a parallel file. Each of the two distributions may be physical or logical. The algorithms are optimized for multidimensional array partitions. We motivate our approach and present potential utilizations. We compare and contrast with related work. The paper also presents a study case, the employment of mapping functions and redistribution algorithms in a parallel file system