As a new computing model, cloud computing has been growing rapidly. At the same time, it brings a serious problem about energy consumption. This paper studies the energy optimization in cloud computing system.High energy consumption in cloud computing is caused by two reasons: first, a large number of servers are idle; the second is the unreasonable task scheduling strategy. This paper proposes a task scheduling strategy-first scheduling with the minimum energy (FSME). The strategy firstly schedules the tasks to server which is in working states and has the minimum energy consumption. The strategy also switches the idle servers to sleep mode to reduce idle energy consumption and reduces energy consumption caused by mode switch of servers. Stochastic Petri net was used to model the cloud computing system and analyze the energy consumption and the performance. Simulation results show that the FSME can improve the energy efficiency while has less bad effect on the performance of system.The server in cloud computing system always running several tasks simultaneously, and when a task is finished the computing resource it used will be idle which will cause energy waste. Based on dynamic voltage and frequency scaling, the paper proposes computing resources dynamic power-aware (CRDP) strategy. When a task is finished and the computing resource it used is idle, with respond time constraint, the strategy reasonably switches the voltage mode of cpu down to low level to reduce its executing power. Then we analyze the energy consumption and the performance of the cloud computing system, and build energy consumption model and performance model. Simulation results show that the CRDP can improve the energy efficiency while meeting the quality of service requirement. |