| 摘要 |
<p>The present invention relates to a load balancing system for a Hadoop MapReduce under a virtual environment. More specifically, a load balancing method includes the steps of: receiving, by a plurality of virtual machines, the times at which tasks have been finished in accordance to a plurality of slots from a virtualization platform; calculating, by the virtual machine, a time remaining until the tasks for all the slots in the virtual machine are completely performed from a preset time; transmitting the remaining time calculated by the virtual machine to a master node; calculating an average value of a plurality of remaining times received by the master node; and performing a control operation such that the master node adjusts a CPU resource allocation amount of each virtual machine in accordance to the average value. Through the above structure, in accordance to the load balancing method and the load balancing system for the Hadoop MapReduce under virtual environment, when the virtual machines performs the tasks, the same time is taken to perform the tasks such that the improvement for performance is able to be expected under the cluster environment.</p> |
