Methodology for Performance Improvement of Future Request Predicting Disk Scheduler for Virtualization

Abstract

Virtualization enables to switch different operating systems without reboot. It enables live migration from one Operating System (OS) to another and results in proportional sharing of storage resources. Virtualization is gaining importance day by day in the fields of academics, industry and business. Performance is the major requirement to fulfill today’s need. As far as, computer’s workload is concerned, there is a need of high performance computing system. As the use of virtualization has increased tremendously there is much focus on optimizing the virtual machine performance. Disk scheduling within the virtual environment plays a key role in optimizing the overall system performance. Prior works on disk scheduling in virtual environment found it difficult to achieve system performance because of the high disk seek time. This paper presents an approach towards the performance improvement of disk scheduling in virtualized environment by future request arrival prediction. The basic idea is to examine whether the traditional High Throughput Token Bucket Disk Scheduling algorithm (HTBS) is still efficient for the performance improvement in virtualized environment.