High-coverage fault tolerance in real-time systems based on point-to-point communication

The distributed recovery block (DRB) scheme is a widely applicable approach for realizing both hardware and software fault tolerance in real-time distributed and parallel computer systems. One of the most important extensions of the DRB scheme which has been outlined in recent years (but not developed fully) is the integration of the DRB scheme and a network surveillance (NS) scheme. We have developed an NS scheme that is effective in a variety of point-to-point networks, called the supervisor-based NS (SNS) scheme. In this paper, we present an integration of the DRB scheme with the SNS scheme, called the DRB/SNS scheme. This scheme is a significant improvement over the previous versions of the DRB scheme with respect to the fault coverage and recovery time bound achieved in those systems that are based on point-to-point networks. The execution support for the integrated scheme has been implemented as a part of the DREAM kernel prototype, a timeliness-guaranteed operating system kernel developed at the University of California, Irvine. The recovery time bound of the DRB/SNS scheme is analyzed on the basis of the prototype implementation.