This paper presents four decades of history and advancement, in high-availability systems for business computing, by IBM and HP/Tandem. The paper chronicles the evolution of IBM’s zSeries, and HP’s NonStop server, from fault-tolerant and high-availability systems, to present-day, 24-7 continuous availability systems.
The zSeries servers are built around proprietary microprocessors with built-in error control, whereas NonStop servers use pairs of commercial microprocessors in lockstep to provide similar error control. Both systems are designed to work correctly after a single failure in hardware or software. The paper presents key similarities and differences in the system architectures of zSeries and NonStop servers. The focus of the paper is on comparing how the two systems implement and provide fault detection and recovery.