Expert Systems for Capacity Management

Expert systems provide a software architecture that can facilitate solving complex problems. This paper describes techniques for, and issues in, expert systems for capacity management, emphasizing expert systems that incorporate analytic models. The paper begins with a brief introduction to capacity management (which consists of performance tuning and capacity planning) and expert systems. Next, expert systems for performance tuning are discussed; three techniques for characterizing performance problems are described in detail: threshold analysis, bottleneck analysis and what's-different analysis. Lastly, we address capacity planning, with emphasis on calibrating predictive models and automating equipment selection.