Method for maintaining data availability after component failure included denying access to others while completing by one of the microprocessor systems an atomic transaction changing a portion of the multiple copies of data

摘要

A method for maintaining the overall system availability of a multi- processor data processing system in the event of a failure at one of a plurality of independent failure points. The system includes a plurality of virtual memory type processor units, each of which may include an interactive terminal, and a main memory which has access to a secondary storage device which is also accessible by the main memory of one of the other processor units. The two units are also directly interconnected by a communications mechanism making it possible to provide a shared virtual memory system. Both processor units employ the same operating system and share the same virtual address space for storing information. The interactive terminals are each connected to more than one processor unit. Failure points include the terminals, the processor units, the communication links, and the various software components that are employed by the system. The method maintains system availability by managing the storage of selected information at preestablished precise points in the processing operation. The method updates identical operating system data structures that are stored in each processor unit with the results of certain identified data processing transactions which have affected information required for succeeding processing operations in manner to insure that either identical updates occur or no update occurs. The method further insures that any changes that might have occurred in the information stored in the data structure prior to the end of an aborted transaction is returned to the initial state to permit the transaction to be retried whenever the path of the falilure can be bypassed.