Main Article Content
Background Cloud and its presence are ubiquitous more than ever after pandemic. Various domain has migrated from on-premises applications to cloud. Due to the increase in cloud adoption, infrastructure management has gained importance multi fold. Infrastructure refers to the servers in the datacenters. The servers are called as Physical Machines (PM) which is composed of logical cores called Virtual Machines (VM). Ultimately jobs are submitted to VMs so VM allocation one of the important tasks in job scheduling. Number of algorithms have poured into cloud market borrowed from various discipline such as nature, mathematics, physics etc. for VM allocation. The proposes approach of algorithms are of hybrid metaheuristic taken from mathematics, physics namely Intelligent Backtracking and Simulated Annealing respectively. Experimental studies show the parameters of concern such as execution time, Energy consumption and cost incurred are optimal when compared to the existing statistical approaches. There is a 48% decrease in the energy consumption and a 90% decline in cost of using the approach. Objectives: The objective of the research work is to bring down the Execution Time, Energy Consumption and cost incurred thereby obtaining an optimal solution for the VM allocation problem in cloud computing.
Methods: The proposed approaches namely Intelligent Backtracking for VM Placement and Simulated Annealing for VM migration are implemented with the parameters to study and analyze the working and feasibility of it. From the related works studied, the parameters of influential importance are Execution time, Energy consumption and cost incurred. Along with it, in the proposed approach Number of Hosts shutdown is also taken into consideration as it has an impact on time and energy parameters. These three parameters affect the cost parameter in a positive way. Conclusions: The proposed algorithm decreases the Mean and standard deviation Execution Time by 21.5% & 33.5% respectively. The proposed algorithm decreases the Number of Hosts shut down by 19%. As a result, the proposed algorithm’s energy consumption is 68.30kWh and improves it by 48%. The research work concludes that hybrid metaheuristic approach on Virtual Machine Allocation using Intelligent Backtracking and Simulated annealing brings optimal results than the statistical, and simple metaheuristic algorithms proposed before. Future work will bring machine learning algorithms for load prediction and devising a new ML based approach for VM allocation in cloud computing.