Generation of pseudo-random test cases

Due to the complexity of large computer systems, the detection of certain system errors may become realizable only after certain input combinations are applied to the system under test (SUT). The state space of a large-scale computer may render the number of different input/ouput combinations astronomical such that, in a deterministic testing, certain effective input combinations are not applied to the SUT (which diminishes the probability of detecting system errors). One of the main advantages of biased random testing is that subtle system input/output combinations can be formed ingenuously. An effective random test generation tool called Systems Assurance Kernel (SAK), which is widely used for testing IBM's large computers such as the Enterprise Systems Architecture (ESA/390), is described. SAK stresses the SUT under different environments while taking into consideration machine features that are architecturally transparent such as superscaler feature.