E-PowerGA: A Genetic Meta-Heuristic for Energy-Efficient and SLA aware Virtual Machine Placement
TL;DR
A genetic meta-heuristic approach to optimize VM placement by consolidating VMs into a minimum number of physical machines in a cloud data center, referred to as E-PowerGA, which significantly outperforms traditional heuristics like Power-Aware Best Fit Decreasing.
Abstract
Cloud computing has emerged as a ubiquitous paradigm for providing computing and storage services over the Internet. With ever-increasing demand for cloud-based services, cloud service providers need to enhance the performance, reliability, and availability of data centers while reducing operational costs. Virtualization is an effective enabler for resource optimization with reduced power consumption; however, efficient VM placement remains one of the key challenges. The paper proposes a genetic meta-heuristic approach, referred to as E-PowerGA, to optimize VM placement by consolidating VMs into a minimum number of physical machines in a cloud data center. E-PowerGA aims at minimizing energy consumption and SLA violations while considering the migration overhead of VMs. Intensive simulations based on real workload traces from PlanetLab illustrate that E-PowerGA significantly outperforms traditional heuristics like Power-Aware Best Fit Decreasing. Experimental results have shown a reduction of 31.44% in energy consumption, 47.31% fewer VM migrations, and 60% improvement in SLA violations during the evaluation period of ten days. These findings pinpoint the efficiency of the genetic meta-heuristic optimization in enhancing energy efficiency and SLA compliance within a cloud data center