Simulation based complex information system capacity planning

The need and value of performance estimation and capacity planning for complex mainframe and client/server information systems is well established and robustly accepted. This paper describes a capacity planning system that has been used by IBM to provide these services to clients for many years. The central component of this system is a discrete simulator with the capability of imitating the behavior and estimating performance characteristics of proposed and/or evolving enterprise information processing systems. This paper presents some of the factors and events that precipitated the development of the underlying simulation language, and precursors of the current model. In order to maintain currency with advances in computer system hardware and software technology, and remain a viable, cost effective, and unique offering, this capacity planning system has likewise evolved. Key elements of this evolution are also discussed. An overview of the simulation model is given to demonstrate its scope and capabilities. Examples are presented that illustrate the type of systems definition data required as input, as well as the performance indicators produced by the model and its relevant reports. Activities related to calibration and validation of the model are also discussed. A workshop format is presented that has been developed and implemented to provide a framework for exercising the capacity planning system.