|Java TM Look and Feel Design Guidelines: Advanced Topics > Part II: Special Topics > 6: Responsiveness > Measuring Response Delays||
An application's user interface must respond in real time. To measure objectively how quickly your application responds, you need to measure its response delays. This section describes techniques for quantitatively measuring your application's response delays.
Measurements of response delays are useful only in relation to benchmark s. A benchmark is a goal that you devise to determine whether your application provides acceptable response delays for a specific task. Without benchmarks, you cannot know for sure whether your application is responsive enough.
Establish benchmarks early in your project by reaching a consensus with representative users and your development team--including management, marketing, and engineers. Your goals for acceptable response delays should be reachable on the minimum computer system that your application supports.
One tool for objectively measuring response delays is a stopwatch. As Wilson and Kesselman explain in Java Platform Performance, testing with a stopwatch has advantages and disadvantages. It is easy to use but is hard to use accurately. In addition, testing with a stopwatch is hard to automate.
A stopwatch is inadequate for measuring milliseconds--which you need to measure when complying with the guidelines in the section Perceived Performance, or Responsiveness. Typically, measuring response delays to the millisecond requires that engineers include tests in the source code of their applications. For text and code examples describing this technique, see the book Java Platform Performance.
|Java Look and Feel Design Guidelines: Advanced Topics.
Copyright 2001. Sun Microsystems, Inc. All Rights Reserved.