Distributed fault tolerance specification through the use of interface definitions

Fault tolerance can no longer be viewed as an exorbitant optional extra, but must be treated as an essential requirement in any distributed system. It is fair to say that if any distributed system is to operate for any contiguous interval of time without failure, a fault tolerant mechanism must be used. This paper reviews and discusses what extensions and functionality may be added to an interface definition language to allow for the description of four different fault tolerant mechanisms.