Method and system for automatic load balancing of advertised services by service information propagation based on user on-demand requests

Abstract

A methodology for balancing demand for networked services in a distributed data processing system is presented. Each client is uniquely associated with a local service manager; one or more local service managers are located throughout a distributed data processing system, and each local service manager provides access to networked services for associated clients. Each local service manager is uniquely associated with a distributed service manager; one or more distributed service managers are located throughout the distributed data processing system, and each distributed service manager provides access to networked services for associated local service managers. A client sends a service request to its local service manager, which returns information about a matching service to the client after finding a matching service that has characteristics that match parameters in the request. If the local service manager does not have information about a matching service, then the request is forwarded to its associated distributed service manager. If the distributed service manager does not have information about a matching service, then the request is broadcast to all distributed service managers. If the distributed service manager has two or more matching services, then it performs a load balancing operation to select a best service to be returned.