Load Balancing Model Based On Prediction Mechanism

The load balancing is an important technique which can improve the availability and scalability of the system performance through the dynamic allocation mechanism. Therefore, it is currently one of the core contents of prediction-based load balancing algorithm that promptly adjust the load balancing algorithm according to each server's workload of the web server cluster.Workload characteristics have an important impact on the performance of load balancing scheduling algorithms in Web server cluster systems. According to analyzing and discussing the role of load characteristics for scheduling algorithm, a prediction-based adaptive load balancing algorithm (RR_MMMCS-A-P) is proposed in this paper. RR_MMMCS-A-P can predict the arrival rate and the size of the follow-up request by monitoring the workload characteristics and rapid adjustment of the corresponding parameters in order to balance the load between servers. Experiments show that compared with CPU-based and CPU-memory based scheduling algorithm, RR_MMMCS-A-P have better performance in reducing average response time for both calculation-intensive and data-intensive jobs.Multi-dimensional Markov chain and queuing theory algorithms can predict the workload characteristics, according to more memory-request more CPU Slice based on Round Robin (RR-MMMCS) mechanism, a Markov forecasting based and queuing models based load balancing algorithm (RR-MMMCS_Markov) is proposed in this paper. Experiments show that the algorithm has better performance in balancing the workload between the individual servers and reducing average response time for each server.