Energy-efficient Scheduling Algorithm For Virtual Machines With Respect To Laod Balancing

With the rapid development of software and hardware technology of computer and the Internet technology, the cloud computing technology has been applied in various applications. Lots of different kinds and sizes of cloud data centers have been built around the world. However, the large number of cloud data centers brings a very serious energy consumption problem to the world. Therefore, developing reasonable and good energy-efficient algorithms plays rather an important role in saving the energy consumption of data centers, saving operating costs and protecting environment.Based on the task allocation mechanisms of data center source dispatching system and the accurate energy consumption models of cloud data centers provided by the GreenCloud simulating system, we proposed an offline energy-efficient dispatching algorithm with delay and an on-line energy-efficient dispatching algorithm with both delay and migration. Both dispatching algorithms employ the delay technique to decrease the servers used for saving the energy consumption of the center. As for the on-line dispatching system, when there are fewer tasks, migration technology is used to migrat the tasks from low-used servers to higher-used servers to decrease the number of servers in use. Meanwhile, keeping load balanced, one server will not be given tasks any more if the server’s load is lighter than 70%.A comparison between the proposed algorithms and other four energy-efficient methods is conducted. All the algorithms are tested in the total energy of the data center, the whole number of used servers, the total running time of all the used servers, the total running time of the dispatching algorithm and the number of tasks rejected by the algorithm. The experimental results demonstrate that the two proposed algorithms have the best performance, which use the least energy, need the fewest servers and the shortest running time of servers and cost the fewest computing sources. What’s more, the proposed algorithms also keep the load balanced for the whole data center.