Analysis of timeout mechanism in a testing-repair model

With a dispatching rule collaborated, timeout is an effort to maximise system performance, by disabling the access of a job to a resource when the job keeps the resource idle too long and then by giving a waiting job the access to the resource. This paper first defines a generic testing-repair model with a single tester and proves that there exists at least one optimal solution in two extreme timeout thresholds with respect to the makespan. Experimental results are presented to support the proof. This problem is then formulated as a discrete multi-objective optimisation problem with five common performance measures of production systems. The two extreme timeout thresholds are efficient solutions in all cases, and they are even dominant in several cases. This paper explores the properties of the timeout collaboration protocol and provides useful insights into applying the timeout mechanism to production systems.