GA-Based Load Balancing in Cloud with OS-level Virtualization

Abstract

The main idea of cloud computing is to provide different computing services to the end-user online over the internet. Cloud computing uses virtualization technology to offer various services to the user. There are two different ways of virtualization; hardware virtualization and software virtualization. The hardware virtualization uses a hypervisor to create several virtual machines from one single physical machine, and software virtualization uses a container engine to create several containers. The traditional virtual machine-based virtualization techniques use more computing resources than OS-level virtualization. Efficient resource management plays an essential role in the overall performance of the cloud system. Evenly distributing the workload among all the available machines is one way of efficient resource management. This paper proposes a metaheuristic approach for load balancing in a cloud system that uses OS-level virtualization. Load balancing is the process of distributing workload or users' tasks equally among all available systems. We implemented several algorithms such as first fit, next fit, best fit, first fit decreasing, best fit decreasing and compared their result. The experimental result shows that the best fit decreasing outperforms all other approaches. Then, we compared the result with our proposed GA-Based approach and found that our algorithm performs better than best fit decreasing.