| Summary: | 碩士 === 明新科技大學 === 資訊管理系碩士班 === 104 === In today's age of the Information technology have growing rapidly, and the computing capabilities of the computer are becoming more powerful. In cloud computing environment, the system how to schedule tasks assign to the resource entities that will directly impact to system performance. In addition, the task's resource requirements and the computing capabilities of the resource entities are different. Past most research focus on viewpoint of task execution or completion time. Past traditional scheduling algorithm may not be valid assign tasks to the appropriate resource entities to execute, that cause some resource entities is busy or idle. If the tasks and resources mapping is inappropriate, it will cause in overall system efficiency is not good, which makes overall system all resource entities load unbalance.
In this thesis, we use another viewpoint of task scheduling with considering the load distribution in cloud environment during the task scheduling process. In our proposed mechanism, two basic steps to schedule tasks including task priority calculation and resource entities selection to perform task selection. In order to provide better cloud resource searching performance, we need a better performance of searching algorithm to find the resource entities in large scale cloud computing environment. We use AVL binary tree architecture to index the resource entities in O(log n) complexity. The basic concept is the maximum request resources's task of the cloud environment have higher execution priority. And the scheduler selects the resource entity that the task's request load is closest to the current overall system average load and assign to the resource entities. And the experimental results showed that our proposed mechanism can solve the task scheduling load unbalanced in the cloud computing environment.
|
|---|
