Adaptable fault tolerance for distributed process control using exclusively standard components

Abstract

This paper describes an adaptable fault tolerance architecture for distributed process control which uses exclusively standard hardware, standard system software and standard protocols. It offers a quick and low cost solution to provide non-safety critical, technical facilities and plants with continuous service. Thereby a maximum of practicability for the application engineers is achieved. The architecture is composed from well known fault tolerance methods under the constraints of real-time requirements. The latitude of non-safety critical applications is carefully used to minimize the fault tolerance overhead. Because of the transparency of the fault tolerance each functional part of the process control, which is represented by an application task, can be implemented without regard to non-determinism and executing hosts. The configuration of a control system is easy and simply done by naming hosts, tasks and groups in a file, wherein every individual task has to be declared with the selected fault tolerance strategy.

It can be expected by a fault-tolerant system that reconfiguration, following a fault, is done automatically. The present system does more: it reintegrates repaired hosts automatically and re-establishes the redundant operation, while the entire system is working.