In Service-Oriented Architecture (SOA)

to improve the adaptability and predictability of the current load balancing algorithms in Web server cluster systems

a prediction based adaptive load balancing algorithm was proposed in this paper.Through monitoring the service characteristics and the arrival rate of workload

the algorithm can automatically adjust the corresponding workload parameters and predict the size and distribution of the follow-up request so as to balance the workload between cluster servers.The study results indicate that the algorithm can effectively reduce average response time in access-intensive distributed Web server cluster systems.