Research on load balancing scheduling algorithm for server clusters based on dynamic resource prediction

In distributed computing systems, task scheduling is crucial for resource utilization optimized and task executed normally. This paper proposes an effective dynamic scheduling algorithm through integrating task resource prediction and real-time server monitoring. The algorithm uses a decision tree model to predict the utilization of task resources and a central node to achieve the real-time monitoring of server clusters so that it can effectively allocate tasks based on the predicted results and the monitoring data. It effectively avoids the issues caused by peak task conflicts which are system downtime, task processing volume decreased and so on. Experimental results show that the proposed algorithm can significantly reduce the average task processing time and remarkably improve the resource utilization rate of the server cluster, which prove its significant advantages of overall system performance outperforms other algorithms.