Abstract: A method and an apparatus for providing scalable layers of highly available applications using loosely coupled commercially available computers. The software running on the loosely coupled computers is divided into three layers: the system layer, the platform layer, and the application layer, each having its own process group activation and fault recovery strategy. A process group contains software processes that depend upon a set of resources common to the process group. In addition to depending upon a common set of resources, processes within a process group share a fault recovery strategy. Fault recovery is performed at the process group level, such that if one process within a process group fails, fault recovery is takes place for all processes within the process group. In the preferred embodiment, an application layer process group may be paired with another application layer process group on a separate computer.

Abstract: A highly available processor complex that can use industry standard hardware and software through use of an N+K sparing arrangement with a separate monitor ("watchdog") processor. The watchdog processor monitors periodic "heartbeat" messages from the monitored elements and can change activity and status of such monitored elements through control messages. Further, the watchdog processor is operatively connected to a power control circuit that controls the power to the monitored elements. Thus, the watchdog processor can guarantee that a faulty element can be rendered non-functional and can also totally restart an element through power cycling it.

Abstract: Platform and/or application software on all online, machine/servers in a cluster is updated without manually taking each machine/server offline. Initially, platform and/or application software for updating is stored in respective directories in an APPLY phase. Next, the new platform and/or application software is activated with or without a trial/test phase in an ACTIVATE phase. Where the new platform and/or application software is activated with a trial/test phase, a ROLLBACK phase is either automatically or manually invoked by the application in the event of a failure of the new software for backing out the new platform and/or application software and reactivating the previous platform and/or application software. An OFFICLAL phase then transitions the new platform and/or application software to the official state followed by a REACTIVATE phase for reactivating the backup copy of the previous/old platform and/or application software after the new software has been made official.