Abstract
Capacity modeling has long been viewed as the key task in estimating future computing needs. Yet the work load estimates that form a model's underlying assumptions are often completely unreliable. One way to improve capacity planning is to optimize input and focus on results analysis.