MIDAS - INTEGRATED DESIGN AND SIMULATION OF DISTRIBUTED SYSTEMS

Iterative transformation of performance models into operational systems is a desirable approach for designing performance-critical systems. This paper describes an approach called MIDAS that supports the design of distributed systems via iterative refinement of hybrid models. A hybrid model is a partially implemented design where some components exist as simulation models and others as operational subsystems. It is an executable model and may be used to determine the stochastic performance characteristics of a partially elaborated design. MIDAS enhances the applicability of hybrid models in system design with its support for interrupts and its inclusion of distributed components in the partially implemented design. The paper describes how an existing simulation language may be extended to program hybrid models, and shows how simulation algorithms may be adapted to execute hybrid models. A prototype MIDAS implementation is operational and was used to develop a set of applications. The experimental results of the exercise are also described.