Detecting Workload Hotspots and Dynamic Provisioning of Virtual Machines in Clouds

Abstract

One of the primary goals of Cloud Computing is to provide reliable QoS. The users of the cloud applications may access their applications from any Region. The cloud infrastructure must be Elastic enough to improve the QoS requirements. In order to provide reliable QoS, the cloud infrastructure must be able to detect the potential workload hotspots for various cloud applications across Regions and take appropriate measures. This paper presents an approach to detect workload hotspots using application access pattern based method in the cloud. This paper also presents how the existing VDN based Virtual Machine provisioning approach [1] can be used to provision new Virtual Appliances at the detected hotspots dynamically and efficiently at the potential hotspots to improve the QoS.