Enhancing performance in a parallel file system

Abstract

As computer applications have become more sophisticated, they have become rather data intensive. Such applications suffer from inadequate use of parallelism for processing data stored on secondary storage devices. Devices such as database machines are useful in some applications, but many applications are too small or specialized to make use of database machine technology.

To bridge this gap, we have introduced a parallel file system. The parallel file system is capable of acting as either an SIMD machine or an MIMD machine depending on the file type.

In the present work, we describe two approaches to enhance performance in the parallel file system. First, we examine a strategy of initiating partial searches when full searches for concurrent file usage are not possible. As a second level, a relocation algorithm has been designed to move selected portions of files (subfiles) to improve the degree of parallelism in the multiprogramming environment. The relocation algorithm makes use of a cost function based on the level of sharing between files to determine the best place to relocate the subfiles.