| With the increasing popularity of Internet services and cloud computing, both energy efficiency and performance guarantees are major concerns for data center operators. However, only a few of previous works elaborate the dynamic system working procedure while using specific mathematical methods to study on the relationship between the two characters.Therefore, this paper builds a mathematical model based on queueing theory to analyze the relationship between the energy cost and performance of a data center. Moreover, we build a optimization objective aimed at improving both performance and energy saving in a data center by finding a tradeoff.The study on the data center in this paper almost includes all the data center issue types, such as the normal model M/M/n and a rare model G/M/A, independent optimization in one single data center and the cooperative optimization scheme among multiple data centers. Besides, we study a lot about the homogeneous data centers and heterogeneous data centers, and we propose different optimization schemes for both of them according to their varying features. Although abstract and too ideal, the dynamic scheduling on homogeneous servers is where this paper started and is able to show some valuable points besides the great difference to heterogeneous models. There are also some good conclusions gained through the simulation over homogeneous data center, which can become very advisable. As for the multiple data center situation, we still adopt the abstract model with no consideration to other elements but load balance and energy saving. We also separate the scheduling decision in the data center from the job allocation decisions among multiple centers in order to simplify the solution. The simulation shows that the cooperation among data centers can make better profits. Finally, according to some conclusions we drew from homogeneous model, we improve a energy saving scheme in heterogeneous data centers by paying more attention in stability. |
PDF Full Text Request