Dynamic Programming Inspired Virtual Machine Instances Allocation in Cloud Computing

Abstract

Cloud computing provides virtual machines instances to the user for performing various computational tasks on demand for a specific period of time. Considering the architecture and characteristics of the cloud environments, traditional virtual machine instances allocation algorithms cannot be applied to the cloud environment appropriately. In this paper, we propose Dynamic programming inspired virtual machine instances allocation algorithm which allocates virtual machine instances to the user based on demand. The aim of this algorithm is to maximize the cloud provider’s revenue. We have mainly focused on the total revenue generation of the cloud provider and percentage of user served rather than focusing on running time and space complexity of the virtual machine instances allocation problem. We evaluate the proposed mechanism by performing simulations. The experimental results show that the proposed Dynamic programming inspired virtual machine instances allocation method provided a higher revenue generation for the cloud provider than traditional fixed price and combinatorial auction greedy virtual machine instances allocation method.