Reliability issues with multiprocessor distributed database systems: a case study

The results and the methodology used in the reliability analysis of a call-handling database system are presented. Those areas where the reliability modeling methods and analysis techniques need to be extended to cover all causes of outages are identified, especially when dealing with novel architectures, such as the next-generation switch. The analysis indicates that for this architecture, node configurations with one redundant processor meet the hardware-allocated downtime objective. However, some configurations do not meet this objective when hardware failure rates, repair times, and other parameters such as uncoverage are increased. Some of these parameters can be controlled by the maintenance policies of the support organization, but the critical ones are mainly defined by the reliability and quality programs that equipment suppliers have in place. Reliability and quality analyses can be used to address the many areas in a supplier´s program that affect product reliability