The computer pool and Fault Containment Regions (FCR)

Fault Tolerance can be obtained by the interconnection of multiple computer systems, called Fault Containment Regions (FCR’s) to a computer pool.

The computer pool must be synchronised by a non priority fault tolerant clock and a self adaptable majority voting algorithm, which must be executed on each FCR. Such a computer pool is specified by the Byzantine theory of Fault Tolerance.

The described computer pool technology, specified by the Byzantine theory, guaranties data consistency in the majority of non faulty FCR’s in the presents of a fault. This is an essential requirement for a fault tolerant computer system.

The fault tolerant computer pool supports the safety and reliability requirements simultaneously due to its scalability feature. An online re-integration function of a recovered FCR allows for on-line upgrading of the redundancy of the computer pool. This is an essential feature for a very long time operability without service interruption.

The fault tolerant computer pool is bus master on a multiple redundant I/O bus-system. The allocation of the FCR’s to the redundant I/O bus system is table driven. A re-allocation of the busses will be executed in case of a computer pool redundancy reconfiguration (degrease, upgrade). However, in each computer pool redundancy configuration, all redundant I/O busses are active controlled by the remaining, operational FCR’s.