Dynamic Resource Scheduling Cloud using Enhanced Queuing Model

Abstract

The important goal of cloud computing is to offer larger data center that satisfies the storage requirements of the customer. The entire data can’t be saved in a single server. Cloud provider (CP) has cluster of servers to fulfill the cloud request from various real time applications. The data is fragmented in multiple servers to maintain availability. Since the data request of a customer needs data from various servers, there is a possibility of attaining dead lock. In this paper, an enhanced queuing model is proposed where the cloud request (CR) is received in queuing manner for allocation of resources. A session is created for the CR with the CP resource allocation from cloud severs. This enables to put constraint on the number of CR making a session with CP to avoid resource suppression. The Wait for Resource algorithm is used for allocating the server resources to a CR without deadlock in a session. This enables to forecast the resource requirements prior to resource allocation phase in a session. This makes the dynamic resource allocation efficient and free of deadlock. The results obtained evaluates the proposed model and helps the CP in dynamically choosing the number of server nodes necessary to achieve better performance for an real time application.