VM Consolidation for Cloud Data Center using Median based Threshold Approach

Abstract

Cloud computing is an on demand computing model which requires large amount of physical devices and provide services to users on the basis of pay per usage model, therefore excessive demand of cloud computing have also led to the growth of computational power inside datacenters. These datacenters consumes huge amount of energy which results high carbon emission. For the optimization of resources and reduction of energy consumption, virtual machine consolidation can be used by switching the idle nodes to sleep mode or by turning them off and by using live migration of virtual machines. Here, we propose a novel method for consolidation of virtual machines such that it meets Service Level Agreements (SLA) and deals with energy-performance trade-off. Therefore, reduction of SLA violation and minimize the performance degradation during migration are two main objectives in this paper. For the allocation and reallocation of virtual resources depending upon their load, this threshold based approach can be used, in which Median method is used to find lower and upper threshold values. Proposed Median based threshold approach is implemented by using CloudSim and validation of this approach is performed across different workload traces of PlanetLab servers and using some random configuration of Datacenters. Experimental results show that this scheme can provide better SLA performance.