Load balance in a new group communication system for the WAN

For widely support the group communication in the WAN environment, a new group communication system CSCWGroup is proposed in the paper. The client/server cluster architecture is introduced in the system. In order to balancing the loads on servers, a novel method for characterizing the workload is presented to replace the traditional workload descriptor - the CPU queue length because of its variability in the stable load state. Vie combination of the CPU queue length and other five workload indices are used to determine the load levels. A modified probabilistic neural network is adapted to classify the servers load states into six types with the indices combination as the input vector. The experiment proves that shorter mean response time can be obtained in the new method. Based on it, load balancing policy is executed in the system to balance loads on servers.