Fault tolerant concurrent C: a tool for writing fault tolerant distributed programs

Abstract

Concurrent C is a superset of C that provides parallel programming facilities. The authors' local area network (LAN) multiprocessor implementation has led them to explore the design and implementation of a fault-tolerant version of Concurrent C called FT Concurrent C. FT Concurrent C allows the programmer to replicate critical processes. A program continues to operate with full functionality as long as at least one of the copies of a replicated process is operational and accessible. As far as the user is concerned, interacting with a replicated process is the same as interactive with an ordinary process. FT Concurrent C also provides facilities for notification upon process termination, detecting processor failure during process interaction and automatically terminating orphan processes. The authors discuss the different approaches to fault tolerance, describe the considerations in the design of FT Concurrent C, and present a programming example.<>