Language support for atomic behavior in distributed real-time systems

We have developed and implemented the RTC language constructs and run-time system to support both timing constraints and shared resource consistency constraints in distributed real-time programs. This paper describes how the RTC language constructs can be used to support timed atomic behavior to maintain consistency in distributed real-time systems that have the possibility of faults. That is, we show how to enforce the requirement that either all actions execute within timing constraints, no actions execute, or there is an exception that indicates that recovery should be performed. We describe RTC support for timed atomic behavior of a single process, timed atomic commitment among distributed participants, and recovery from constraint violations due to faults.<>