Preemptive Task Scheduling Algorithm Based On Availability In Heterogeneous Systems

In the past few decades, heterogeneous system has been widely used in science and business. In recent years, many scholars have focused on the scheduling algorithm in heterogeneous systems in order to improve application performance. Scheduling theory of the basic assumption is that all machines are always available to handle the task. This assumption may be reasonable in some cases. But it is not effective when there are some kind of maintenance requirements ,interruption or other restrictions which the machine can not deal with , these machines can not be used to handle the task. In fact these constraint factors may be in many applications.In this article, availability is defined as a computing node in a given time interval running time of the total proportion of the time. Now many high-performance applications require high-availability computing platform, such as military applications, medical applications , the international business applications and so on ,require very high availability of services, as long as there is a computing node is not available may cause serious or fatal fault. Therefore, in order to deal with accidents and maintenance activities, scheduling strategy in heterogeneous systems must take the availability factor into account.In order to solve these problems, based on the intensive study of the availability in computer systems , we set up the heterogeneous task scheduling model of availability , analyze the feasibility of the model , propose the scheduling problem of preemptive tasks, propose a preemptive scheduling algorithm P-SSAC which is based on availability and more supportive for the priority task, establish a mechanism to detect load balancing , and expand the reality applications of the availability.On the basis of the current algorithm SSAC,the algorithm P-SSAC has similar performance with the existing algorithm, but it can work under the preemptive mode, and can keep a good balance between the availability and the response time to improve the success rate of scheduling..In the simulation experiments, we construct a heterogeneou system with sixteen nodes,compare the experimental results between the new algorithm P-SSAC and some classic methods by making use of the grid simulator GridSim, which shows that P-SSAC algorithm significantly improves the availability of the system, because it takes the availability needs of tasks into account when allocating the tasks to the heterogeneous nodes.