A design and modeling environment to develop real-time, distributed software systems

Abstract

Design of real-time, parallel/distributed systems is a very complex task. First, the system needs to satisfy not only a desired set of functional requirements, but also a set of performance requirements. Second, architecture design becomes an essential issue in the parallel/distributed environment since complexity of the new availability of parallel and distributed processing resource has added a new dimension to the software design process. Our goal aims at integrating performance engineering into software engineering under a parallel/distributed environment. A systematic methodology for constructing parallel/distributed software systems to meet performance objectives, and matching between software structures and architecture configurations is presented as an integral part of the software design process. A Software Performance Analysis and Design Environment (SPADE) for real-time, parallel/distributed software systems is developed to provide a richer environment for the designer to cope with different modeling and analysis methods during the different design stages. It is an iterative development system with feedback information to assist the designer in adjusting previous designs. It captures the dependencies and relationships between different tools and integrates a variety of modeling techniques to represent different aspects of software performance. It provides an ability to make more intelligent design decisions during the design stage.