Fair Fit—A Load Balance Aware VM Placement Algorithm in Cloud Data Centers

Abstract

Cloud computing is a kind of large-scale distributed computing. It has recently emerged as a new technology for providing services over the Internet. It has gained a lot of attention in individual, industry, and academia. Cloud computing dynamically allocates virtual resources as per the demands of users. The rapid increase of data computation and storage in cloud results in uneven distribution of workload on its heterogeneous resources, which may violate SLAs and degrades system performance. Distributing balanced workload over the available hosts is a key challenge in cloud computing environment. VM placement is the process by which it selects the most suitable physical machine (PM) in cloud data centers to deploy newly created virtual machine (VM) during runtime. In this paper, we study VM placement problem with the goal of balanced resource utilization in cloud data centers. Several algorithms have been proposed to find a solution to this problem. Most of the existing algorithms balance the cloud resources based on its utilization which results in unbalanced and inefficient cloud resource utilization. We propose an algorithm, which minimizes the imbalance of resource usage across multiple dimensions, reduces resource leak and wastage, and maximizes the resource utilization. Our algorithm finds a suitable host for incoming VM from categorized host list based on remaining resources using cosine similarity measure. Simulation results show major improvements over the existing algorithms like Binary Search Tree-based Worst Fit and Best Fit, Round Robin, and default Worst Fit.