Improving I/O performance in distributed file systems for flash-based SSDs by access pattern reshaping

Abstract

Cloud computing is being widely adopted in the industry due to improve resource utilization and provide more computation power. In cloud computing systems, many numbers of users execute various types of applications that produce a large amount of data. To store large data, distributed file systems are used in many cloud computing systems. Recently, to improve the performance of distributed file systems, emerging flash-based solid-state drives (SSDs) are widely adopted since they provide high performance compared with existing hard disk drives (HDDs). However, due to the characteristics of SSDs, the performance of distributed file systems is greatly impacted by the access patterns from the applications. For example, the performance of random writes in SSDs is lower than that of sequential writes. In this article, we propose an address reshaping technique for SSDs in distributed file systems to improve the performance of random writes by transforming the access patterns. Our scheme reshapes random write requests into sequential write requests in the distributed file systems. This enables the distributed file systems to issue the sequential write requests to SSDs. The experimental results show that the proposed scheme improves performance by up to 46.26% compared with the existing scheme.