An Effective Approach for Virtual Machine Migration and Dynamic Placement Using Elephant Herd Optimization

Abstract

Cloud computing is a platform for offering computational services as a method to deal with multiple problems in virtualized data management. Therefore, it is necessary to position and migration of virtual machines in order to accomplish several contradictory objectives. This work explores the state-of-the-art in the field in regards to the difficulty of these tasks and the vast number of existing proposals. Cloud Measurement combines new technologies that shape our lives in a way that saves investments in the upfront infrastructure for consumers operating on VMs on physics machines provided by a cloud service. Multiple VMs on the same PM could have different work completion times due to the heterogeneity of numerous works. PMs are heterogeneous in the meantime as well. Consequently, multiple VM placements have differing completion periods. Our goal is to reduce the completion time for VM input requests through a realistic schedule for VM placement. This dilemma is NP-hard so it can be simplified to a problem with knapsack. We suggest an offline approach for VM placement by way of emulated VM migration, and an actual migration mechanism for VM solves the online VM placement. The migration algorithm is a heuristic approach, where we explicitly position the VM to its best PM, given that it is capable of doing so. Otherwise we can move another VM from this PM to handle the new VM if the migration limitation is met. In addition, this work incorporates and suggests the introduction of the online dynamic positioning Elephant Herd Optimization (EHO) approach, and the assessment results show the high efficiency of the proposed algorithm.