Friday, September 02, 2005
TTN On-Line Edition for September 2002: "Load Testing Terminology by Scott Stirling (email@example.com) Introduction What is the difference between load, stress and performance testing? Why do these three types of testing seem to belong together, perhaps with others such as scalability testing and benchmarking? Questions such as these, which I have encountered in my own career as sometimes load tester, on Web forums and in discussions in the workplace, are what this article purposes to answer. It would be nice to point to some seminal piece of QA literature that settles these questions definitively, but surprisingly, this is not possible. The use of these terms in the QA literature varies to the point that sometimes load testing is defined as a type of performance test, or performance test is a type of load test, or (somewhat common, actually) load testing is not worth mentioning explicitly. To what degree definitions proposed in the literature have been arrived at independently versus influenced by or directly based on previous published definitions is impossible to tell. The signature characteristic of background, performance and stress testing is that they all require some kind of definite workload (the terms 'workload' and 'load' are interchangeable in this context) exercising the system or component under test (SUT or CUT) during testing. Load is an indispensable part of the test set up for all these types of testing. There are notable exceptions where simulated load is crucial to other types of testing, such as aspects of security testing, where it may be required to simulate load- related conditions giving rise to security problems (such as a buffer overrun or denial-of-service). Reliability testing sometimes requires load as a prerequisite to measuring realistic, time-dependent phenomena such as mean-time- between-failures (MTBF) in transaction or batch processing systems. But reliability is a concern for other criteria, such as functional accuracy "