Systems reliability and availability prediction and comparison with field data for a mid-range computer

The author describes reliability and availability prediction based on systems evaluation through Markov modelling and comparison of results with field observations of a sample of mid-range computer systems with mirrored disks. The system uses single level architecture with disk striping. Reliability evaluation revealed that, with mirroring, the disk system and the intermediate FRUs (field replaceable units) between the disk system and the processor will be duplicated. Hence, system failures would be limited to the processor which is not duplicated. The validity of the prediction was checked by field monitoring of the performance of four systems with mirrored disks. Good agreement between the predicted and observed reliability and availability estimates was observed reliability and availability estimates was observed for the system hardware. As a result of disk monitoring the disk system is totally fault tolerant