Abstract: It is essential to control system actions in the condition that every component functions normally,as well as in the environment that some component gets failed,when designing an architecture of a distributed fault-tolerant system.Designs for the both situations are generally tightly coupled,which increase the difficulty of understanding,designing,developing,and maintaining a distributed fault-tolerant system which may be large-scale and may provide strong fault tolerance.To settle this problem,a new method was proposed in this paper,which defines a description language by the Vienna definition language on the basis of Hoare’s theory of communicating sequential processes.It can describe not only the property of concurrency in distributed computing but also how a system tolerates failure.More importantly,it seems to indicate that programming languages are doomed to be more abstract in higher level in future,and it may provide a new idea in the field of distributed fault-tolerant computing.

Key words: Distributed fault-tolerant computing,FTDL,VDL