A learning automata-based algorithm for energy and SLA efficient consolidation of virtual machines in cloud data centers

摘要

AbstractResource management in cloud computing consists of allocating processing resources, storage, and network to a set of software applications. Resource providers focus on performance and utilization of resources considering the constraints of service level agreement. Resource performance is achieved by virtualization techniques, which share infrastructure of the resource provider between different virtual machines. This study proposes a novel algorithm based on learning automata, which improves resource utilization and reduces energy consumption. The proposed algorithm considers changes in the user demanded resources to predict the PM, which may suffer from overload. Due to preventing server overload, the proposed algorithm improves PMs’ utilization, reduces the number of migrations, and shuts down idle servers to reduce the energy consumption of the data center. The proposed algorithm is simulated in CloudSim simulator; the 10-day processor information of a real PlanetLab cloud infrastructure system are used for workload data. Performance of the proposed algorithm is compared with existing algorithms such as DVFS, NPA, and the threshold algorithm in terms of energy consumption and the number of shut down PMs. Simulation results indicate that the proposed algorithm outperforms other algorithms with 175.48 Kwh, 0.00326 in energy consumption, SLA violation respectively.Highlights•Proposing a Power and SLA efficient resource allocation algorithm optimizing energy consumption, number of VM migrations and SLA violations.•Using Learning Automata to adapt to environment parameters.•Employment of learning automata to prevent of service level agreement violation and physical machine overload.