Method and apparatus for on-demand resource allocation and job management
View Patent ↗The invention is a method and apparatus for on-demand resource planning for unified messaging services. In one embodiment, multiple clients are served by a single system, and existing system resources are allocated among all clients in a manner that optimizes system output and service provider profit without the need to increase system resources. In one embodiment, resource allocation and job scheduling are guided by individual service level agreements between the service provider and the clients that dictate minimum service levels that must be achieved by the system. Jobs are processed in a manner that at least meets the specified service levels, and the benefit or profit derived by the service provider is maximized by prioritizing incoming job requests within the parameters of the specified service levels while meeting the specified service levels. Thus, operation and hardware costs remain substantially unchanged, while system output and profit are maximized.
1. A method for optimizing an allocation of system resources, comprising:
determining, for each source from which a system receives job requests, a minimum percentage of job requests to be processed by the system, where the system receives job requests from at least two separate sources;
allocating system resources among the at least two separate sources to meet at least one job processing obligation and to maximize a total profit of the system; and
scheduling a processing order of new job requests on-line, wherein said scheduling is accomplished by:
determining a completion deadline for a new job request;
determining an estimated process time for the new job request as a length of time needed to process the new job request by the completion deadline; and
scheduling processing for the new job request based on the completion deadline and the estimated process time as compared to completion deadlines and estimated process times for existing jobs in the system;
routing the new job request to a server in the system that can complete processing for the new job request soonest;
assigning the new job request to an end of a job queue of the server;
determining whether a cumulative length of time necessary to process all jobs in the job queue, including the new job request, exceeds a latest completion deadline among all jobs in the job queue; and
temporarily stopping work on a job in the job queue having a longest estimated processing time, if the cumulative length of time necessary to process all jobs in the job queue, including the new job request, exceeds the latest completion deadline among all jobs in the job queue.