Green Scheduling of Bag-of-tasks Applications in Cloud Data Center Using GreenCloud Simulator

Abstract

The cloud provider provisions the resources of the data center to the user so as to satisfy the Quality of Service (QoS) negotiated between them and specified in the Service Level Agreements (SLA). To attain maximum profit, the cloud provider has to schedule the tasks in the data center efficiently in accordance to the users QoS requirements and also minimize the energy consumption thus reducing the operational expense (OPEX) of the data center.  The energy efficient scheduling of cloud applications has to consider the QoS parameter specified as job deadline and assign them to minimum number of servers possible without affecting the performance so that the idle servers can be shutdown to reduce the energy consumed. This paper analyses this green scheduling for various workloads of the cloud users and provide information about the task completion and rejection at various data center load, the SLA violations and the degradation of system performance. It presents the server utilization specifying the number of busy and idle servers with their computational load and the energy consumed by them for different workloads. The paper also provides the insights on the utilization of the servers for the workloads and analyses the reasons for the underutilization of servers and how the scheduling can be improved without sacrificing the performance of the data center. The paper analyses the above observations of the green scheduling of the data center using GreenCloud.