An improved on-line monitoring technique for a fault-tolerant computing node

摘要

A fault tolerant computing node is an indispensable component of reliable distributed computer systems devised for life critical applications. The on-line monitoring technique is frequently used for error detection in such systems and assumes the use of an external hardware monitor connected to the system bus. It does the control flow checking based on signatures assigned to each block of an application. However, this technique cannot be used with contemporary processors with the built-in cache. Therefore, an improved on-line monitoring technique which overcomes this problem is proposed in the paper.