Reduction of Energy Consumption in Cloud Data Centers with Proper Placement of Virtual Machines

In today's world, the use of smart mobile devices for easy access to data and processing resources is expanding rapidly. Rapid technology growth and the increasing number of users make hardware and software architectures upgrade a constant need. The necessary infrastructure to implement this architecture is the use of virtual machines in physical systems. The main issue in this architecture is how to allocate virtual machines to physical machines on the network. Each virtual machine has a set of requirements. The process of allocating resources to virtual machines should be such that any hardware can meet all the needs of virtual machines. We should notice that the allocation pattern of virtual machines directly affects parameters such as load balancing, energy consumption, and cloud throughput. So far, various solutions have been proposed in this area and each of them has a specific goal including symmetric distribution of load, decreasing energy consumption, etc. In this article, a new solution to the optimal allocation of virtual machines in the cloud computing system is presented. We have proposed a method that we call it Learning Automata-based Multi-Objective Cuckoo Search (LAMOCS). The multi-objective cuckoo optimization algorithm is used to achieve an optimal solution. The proposed method simultaneously uses three criteria in order to measure the fitness of each solution: "physical resource utilization rate", "load balancing" and "energy consumption". The proposed method offers a solution with minimum energy consumption, the least rate of cloud resource waste, and better load balancing. LAMOCS uses learning automata as a reinforcement learning model to improve the performance of the optimization algorithm for optimal placement of virtual machines. Also, it helps the search algorithm to converge more quickly to the global optimum. The simulation results show the proposed method has been able to perform about 7% better than the genetic algorithm in reducing energy consumption and also it acts about 8% better than the other compared methods in terms of load balancing.