Research On Dynamic Task Scheduling Algorithm In Grid Computing

The grid is a hot technique developped in recent years, it makes full use of all kinds of available resource which scatter in different geography position to form a super computer. In the future the internet technique will collect all kinds of idle resources such as computing resources, memorizer resources and apparatus resources etc. To draw out an effcient task scheduling alogrism is very important in grid researh. The goal of task scheduling is try its best to match application needs with using available resources reasonably in a dnyamic grid situation. It is needed to slove the difference between host calculation abilities and communication capabilities reasonably, so we can distribute tasks in an optimisticly manner. However finding optimal schedules in grid system has been shown in general to be NP-hard.It is very important and realistic to put forword a better scheduling algorithms based on existing alogrism which can make full use of all kinds resources and improve the computing efficiency.This paper firstly introduce the conception of grid, summarize the status quo and the development trend of grid in recent years,and expatiate the importance of the task schedule in the study of grid. Then thorough analysis and comparisons with regard to various scheduling algorithms, both static and dynamic, are made. Base on this, we present a kind of trust mechanism-based dynamic task scheduling model which imports trust mechanism by using interpersonal trust relationship in human society for reference. This model collects and feeds back resource information through the use of MDS and NWS. It adopts the method of DSRL for the management of data repository and the method of dynamic transfer for the tasks in the trouble nodes. For the task scheduling strategy, the Min-min algorithm is modified and the Trust-Min-min which enhances the validity of grid computing algorithm is proposed. At last we simulate our algorithm with the aid of SimGrid toolkit and verify that our algorithm is reasonable and efficient.