On-Line Real-Time Service Allocation and Scheduling for Distributed Data Centers

With the prosperity of Cluster Computing, Cloud Computing, Grid Computing, and other distributed high performance computing systems, Internet service requests become more and more diverse. The large variety of services plus different Quality of Service (QoS) considerations make it challenging to design effective allocate and scheduling algorithms to satisfy the overall service requirements, especially for distributed systems. In addition, energy consumption issue attracts more and more concerns. In this paper, we study a new energy efficient, profit and penalty aware allocation and scheduling approach for distributed data centers in a multi-electricity-market environment. Our approach efficiently manages computing resources to minimize the processing and transferring energy dollar cost in an electricity price varying environment. Our extensive experimental results show the new approach can significantly cut down the energy consumption dollar cost and achieve higher system´s retained profit.