An Enhanced Approach for Performance Improvement using Hybrid Optimization Algorithm with K-means++ in a Virtualized Environment

Virtualization plays a vital role in cloud computing. It provides better manageability, availability, optimistic provisioning, scalability and resource utilization in current cloud computing environments. However the performance issue is a major concern in virtualization. The performance of the application running inside the virtual machine gets affected by the interference of the co-virtual machines. This approach provides a novel task scheduling mechanism that allocates the suitable resources to virtual machines which are running in parallel. An interference prediction scheme is proposed to utilize characteristics that are collected when an application running on virtual machines to maintain less system overhead. Nelder-mead method is employed in prediction to create relationship model from the observed response and control variables. A hybrid algorithm: Ant Colony Optimization and Cuckoo search algorithm with K-means++ is adopted for task scheduling process. This approach shows effective improvements in terms of throughput and execution time.