Shared Data Availability Across a System Failure
Original Publication Date: 1985-Aug-01
Included in the Prior Art Database: 2005-Feb-19
This invention relates to a method for managing access to a shared resource in the presence of failure among concurrently executing first and second processes resident in respective hosts. Each host includes an associated operating system having a resource lock manager for establishing and maintaining locks to addressable subsets of the resource, said manager reserving access to a requesting process. The method comprises the steps of (a) identifying the addressable subsets of the resource pertinent to the critical sections of each process; (b) communicating said subset identifiers among the lock managers; and (c) upon unavailability of either process, maintaining only those locks of the unavailable process required to preserve its critical sections.