SELF-TUNING VIRTUAL STORAGE MANAGEMENT FOR DEDICATED REAL-TIME COMPUTER SYSTEM

A self-tuning and efficient computer method is disclosed for the management of virtual storage applicable to dedicated real-time computer systems. This method interposes a layer between the application and the real-time computer system to form a two-layered structure to meet an application's request for virtual storage (i.e. buffer request). The method adds a real-time system's slower allocation (second layer) to guarantee the creation of all other buffers during one real-time period. The self-tuning first layer is checked first to satisfy a buffer request, while the untuned second layer is entered to create a buffer when the first layer fails; either the request size is not yet tuned, or the pre-allocated buffers have run out. These entrances to the second layer provide a monitoring mechanism from which a new pre-allocation definition, based on the system usage history, is derived to tune the first layer at the next initialization time. This method also monitors "storage leaks" in which buffer pre-allocations are repeatedly insufficient because some repetitive requests are unmatched by corresponding free requests.