Performance model directed call assignment algorithm in a distributed system of dissimilar processors

In this paper, a new algorithm for scheduling call processing (CP) requests in a distributed system of disparate application processors (AP) connected by any high-speed interconnect is presented. Such systems have applications in wireless communication, automatic call distribution in center applications. The algorithm QR is based on a performance model of the CP scenario within the underlying architecture. It assigns a new CP request to an AP that has the highest promise of completing the task in the least time within the stipulated deadline. It focuses on each CP request in real time, and turns it down only if the system can not handle the call. Further its inability to process a call will have no impact on the decision on the next CP request arrival. The system performance and capacity gains effected by the algorithm are analyzed and illustrated. Thus, in a 2-AP system where one is ten times faster than the other, the QR algorithm processes 17% more busy hour calls (BHC) per second at 1 second mean delay cut-off and 100% more BHC per second at 0.3 second delay cut-off than the existing AP utilization equalization heuristic