Response Time Analysis of Distributed File System Based on Fork-join Queue Model

Abstract

Application performance is criterion for distributed storage system evaluation, and hence the specific task of system configuration and performance tuning for these application is of great interest. The effectiveness of automatic storage management depends on the accuracy of the storage performance models that are used for making resource optimization configuration. Distributed file system plays a key role in mass distributed storage system. Although the potential advantages of distributed file system in application workloads have been documented, the potential impact to application performance in mass distributed storage environments is not clearly understood. This paper presents a response time analysis approach based on fork-join queue model for distributed file system. The performance technique is potentially useful in the early phases of design distributed storage system when several candidate configurations have to be evaluated quickly. The experiments validate our response time analysis model and prove our approach is appropriated to build performance model for distributed storage system.