Load Based Migration Based on Virtualization Using Genetic Algorithm

Abstract

Load balancing is the process of distributing tasks among different nodes in a network. The nodes may be either on different machines or nodes (virtual machines) on the same machine. Based on the availability of the nodes, processes from fully loaded node can be migrated from one node to another having less load known as process migration. Process migration can also be done in the virtualization environment by using a hypervisor called Xen hypervisor. This hypervisor safely multiplexes the hardware resources of the physical machine leading the resource allocation in the Virtual Machine (VM) to improve the utilization and performance. To optimize the task of balancing load among nodes, Genetic algorithm (GA) may be used for selecting the destination. GA is a search algorithm, based on natural genetics and principle of evolution and is been widely used in optimization with binary and continuous variables. With the adaptive crossover operation of GA and the searching heuristic and fitness function, lots of possible solutions are searched and the best one is selected as the destination for migration. GA is been proved to be effective in discovering the global optimum even in a very complex searching space.