Resource optimization and cost reduction by dynamic virtual machine provisioning in cloud

Abstract

Cloud computing delivers platform, software application and hardware infrastructure as a service over Internet. It allows the users to utilize the service on-demand and pay-per-use model as given by Amazon EC2. There are two types of cost involved, fixed and variable. Cost will rise as the demand increases. In this paper the web service request coming from users are categorized into groups and virtual machines (VM's) are created group wise from the physical servers. Mapping of these two group id is done by application provisioner. Fixed cost is charged on requesting reserved resource as per Service Level Agreement (SLA) and variable cost for requesting instantaneous resources. Ultimate cost charged to the user is minimized by 16%. Efficient workload monitoring, grouping of VM's and user request helps to finish user request with in time and better utilization of available resources (physical serves). Idle time of resource is reduced from 50% to 23%.