Automatic Generation of Distributed System Simulations from UML

Nowadays, an object-oriented approach is commonly used for building computer systems. The benefits of the object-oriented method, such as scalability, stability and reusability, make this method suitable for building complex systems, including those in the distributed system area. A distributed system application usually needs to satisfy quite stringent requirements such as reliability, availability, security, etc. and the cost of building such an application will be quite high. It is therefore desirable to be able to predict the performance of the proposed system before the construction begins. In order to do this, it is important to evaluate the requirements of the new system and translate them into a specification (design). The design process helps the system developers to understand the requirements better as well as to avoid misconceptions about the system. From the specification, a simulation program can be built to mimic the execution of the proposed system. The simulation run provides some data about the states of the system and from these data, the performance of the system can be predicted and analysed.

UML (Unified Modeling Language) is one example of the object-oriented design methods that has been widely used for specifying system requirements. There are also some object-oriented simulation languages/packages available, for example, SIMULA or C++SIM package, but it is often difficult to transform the system’ s requirements into a simulation program without sound knowledge of some simulation techniques. On top of that, a new simulation program needs to be built each time for different systems, which can be quite tedious. The currently available UML tools do not provide a feature to generate simulation programs automatically from UML specifications. In this paper, we describe a tool for constructing simulation programs in a generic way, based on a simple specification (preferably in a UML notation) by identifying the simulation components and their structure.