Virtual Machine Consolidation Based on Load Distribution and Realtime Scheduler With Multi-objective Optimization

Abstract

Abstract Optimal resource utilization and reduced energy consumption have been the primary objectives of cloud data centers as the dependency on cloud platforms is increasing day by day. Consolidating the virtual machines is a standard procedure for addressing the common issues and meeting the objectives. Though the approach seems viable for effective functionality, it is observed that consolidation performed over the permissible limit may result in violating the service level agreements in cloud service providers. When energy conservation is concentrated in the cloud platforms, multiple other factors are neglected or compromised. The supposed strategy for effective virtual machine consolidation must contemplate the parameters such as quality of service, service level agreements, reducing violations, resource distribution, load management, migration overheads, network resource management and other communication protocols. The proposed approach focusses on determining the dynamic load and resource management based on multiple objectives in order to reduce the power consumption. The dynamic load is derived based on a time-series analysis over the distributed load in different time zones. Increment in load distribution owing to virtual machine consolidation and selection is observed for improving the efficiency of consolidations. The load prediction approach along with current load detection has included multiple objectives as desired. The proposed approach, from the experimental analysis, has delivered a promising solution for load prediction, distribution and energy conservation in cloud service providers and optimized the functionalities of users. The energy efficiency was observed to be higher than existing virtual machine consolidation approaches along with effective load sequencing and maintaining the service level agreements.