Published Paper Details:

Virtualization is a prime factor in Cloud Computing; there are various virtualization technologies existing in the industry, nowadays virtualization is implemented with good performance benefits. It has achieved, very good overall performance, but it shows deflection in the system and IO performance concerning the workload running on virtual machines. If the VM has CPU intensive Workload then it produces considerable overheads in system performance if VM has a mix (CPU and IO intensive) type of workload, it produces a bit more overhead. However, IO intensive workloads like databases still suffer from performance degradation in virtualization. In this paper, we have analyzed disk IO handling mechanism in para-virtualization technology Xen. We have conducted experiments with database workload measured performance with different IO scheduling strategies and proposed a framework for improving disk I/O performance for database workload in a virtualized environment, which consists of a possible configuration of Linux IO scheduling strategies according to its features. The key idea is to make disk I/O requests handled by the correct scheduler instead of default scheduler shows better performance in IO Wait and read and write operations.