| Summary: | 碩士 === 中國文化大學 === 資訊管理研究所 === 93 === Grid computing architecture had already been defined a complete physical layer. The functions of information systems on Grid architecture are resources (such as computing and storage) sharing, collaborative processing, reliable and secure connection, etc. However, each resource of coordination node in the grid environment, e.g. CPU loading, memory rate of utilization, etc., change dynamically, how make resource these optimization plan and allocation, in order to improve the whole efficiency of the grid environment, it is an extremely important topic.
In this paper, we proposed a dynamic resource analysis model (DRAM) based on grid computing architecture. This model can make resource of coordination node optimization plan and allocation. There are four sub-modules in this model, saying, user interaction model (UIM), job manipulation model (JMM), data analysis module (DAM) and computation analysis module (CAM). UIM received the user’s processing requirements (e.g. data files, or computation, etc.) and transferred the result of execution to user. JMM can monitor status of nodes e.g. storages space, CPU rate of utilization, etc. DAM can divide the data files into some replicas, and store them to different grid nodes. RAM can provide three kinds of different allocation methods, for users to choose depending on requests of processing. Via this model, it can make the resource allocation more efficient, and can decide allocating methods on demands.
We constructed DRAM by Globus Toolkit 3.0.2. Via the result of the experiment, DRAM can make resource allocation have the more efficient and security.
|
|---|
