Development of Hybrid Model For Optimizing Virtual Machine Selection Using Cloud Services

Abstract

Cloud computing provides pool of convenient and on demand computing resources. In newlinethis modern world cloud resources are widely used in most of the industries, providing newlinethe customized services to its users. Classifying and maintaining cloud resources is a newlinegreat challenge. Cloud providers use their in-house resources to create different services newlineto cater to the need of demanding organizations. There are three types of services such as newlineSoftware as a service (Saas), Infrastructure as a service (Iaas) and Platform as a service newline(Paas). These services are highly scalable and provides customized model and allows newlinetheir use via the internet. In our research we mainly concentrate on Infrastructure as a newlineservice (Iaas) Model, because managing this service model is an efficient way to improve newlinethe performance of the system. newlineAccording to the cloud computing paradigm, cloud services make use of Virtual newlineMachines (VMs) that are running on Physical Machines (PMs). newlineA user consumes VMs that are hosted on PMs located at one or several data centers. newlineEnergy consumption of a data center is highly correlated with the number of active PMs. newlineData center relies on a placement algorithm to assign the required VMs on a minimum newlinenumber of PMs in order to minimize energy consumption. In modern age a huge number newlineof different kinds of applications are processed by data centers. These data centers newlineestablishment incur high cost in purchasing IT resources and their maintenance. Cloud newlinecomputing model facilitates creation of extensive scale virtualized data centers with the newlinegoal that clients can utilize them on interest on a compensation as-you-go premise. These newlinedata centers consume unprecedented amount of electrical energy which increases the newlineoverall operating cost and carbon dioxide emission. Energy consumption of cloud data newlinecenters can be reduced by using Virtual Machines (VMs) which optimize their resource newlineusage. newlineVirtual Machine optimization is an effective solution for optimizing the computing newlineresources which can achieve by reducing the below parameters: newline1. Execution time newline2. Number of active resources deployed newline3. Energy consumption newline4. The operational cost of the organization. newline newlineIn our research work we have designed and developed different energy efficient models newlineto optimize resource usage in the cloud datacenter by reducing the number of active newlinePhysical Machines thereby minimizing the energy consumption. newlineThe thesis presents the solution to the problem of optimizing the VM Selection and newlinePlacement. We optimize the selection and placement of VMs in such a way as to newlineminimize the energy consumption and maximize the resource utilization in cloud newlinedatacenter. The optimization process involves two optimization techniques for newline newlineimproving the overall resource optimization in the cloud datacenter. We use a Nature- newlineInspired algorithm Particle Swarm Optimization (PSO) and First-Fit Decreasing (FFD) newline newlinealgorithm to achieve our proposed goal. In the thesis, we have given brief overview of newlineboth the algorithms used and the custom modification done to adapt to the VM placement newlineoptimization. Our hybrid algorithm performs much better than the traditional approaches newlineand the results of traditional and our hybrid algorithm are compared and presented in the newlineresults section. The obtained results indicate that our proposed method performs better in newlineterms of resource wastage, the number of Physical machines used, and in utilizing the newlineCPU cycles when compared with the existing algorithms. The algorithms are analyzed newlineand evaluated using Cloud simulation Toolkit. newline