A Load-balancing method for high performance cluster computing system

Load-balancing is an important technology in high-performance cluster system, and the performance of the system can be improved by adjusting the task load of cluster nodes. Considering the factors of the computer itself and the current load status of the computer, a dynamic Load-balancing method based on capacity evaluation is designed. Through decentralized collection and centralized management of load status information, cluster nodes collect load information when their status changes and report it to the control node. The control node obtains a computer capacity evaluation according to the difference between the maximum processing capacity of the computer itself and the current load capacity of the computer, and forms a capacity evaluation vector table. Finally, this vector table is used as the basis of Load-balancing algorithm. In addition, in order to balance the system load more effectively, service scheduling is used to realize the initial distribution balance of system services. Using task migration to realize the dynamic adjustment of system load. Compared with other existing Load-balancing methods, the proposed dynamic Load-balancing method shortens the service response time, reduces the extra overhead caused by system Load-balancing, and can improve the performance of the cluster.