A fault-tolerant protocol for load balancing system

In this research, an add-on Fault-Tolerant Protocol for Load-Balancing System is proposed. Load Balancing is widely used in the market, however, state recovering of failed job usually is a problem. Duplicating server is usually used to handle this problem in practical situations, however the cost will be increased. Under this protocol, the peer-server group can take over the failed jobs handled by the downed server automatically and the previous states of the connections also can be recovered at the same time. Those processes are totally transparent to the client. The advantage of this protocol is that Load-Balancer does not take place in the job recovery process and no states information is stored in the Load-Balancer. In other words, this protocol can be applied to any job distributing algorithm and minimal modification is required for existed Load-Balancing Systems.