An Adaptive Weighted Least-Load Balancing Algorithm Based on Server Cluster

In order to resolve problems such as server's weight described inaccurately, load balancing schedule inefficiently and so on, which exist in the traditional load balancing algorithms Based on server cluster, a kind of adaptive weighted least-load balancing algorithm is founded in this article. The server nodes are divided into mild load nodes and non-mild load nodes. The servers are polled in the cluster, and the mild load node is selected to response users' requests directly when user-connection requests arrival. If there are no mild load nodes in the cluster, weights are adjusted adaptively according to the server connection status, and the nodes are selected which has the minimum ratio of weight to load to provide service. Through simulation on OPNET Modeler, the results indicate that the new proposed algorithm can reduce the service delay, improve the average response rate, and balance servers' load in the cluster effectively.