An investigation of distributed computing for combinatorial testing

Summary Combinatorial test generation, also called t‐way testing, is the process of generating sets of input parameters for a system under test, by considering interactions between values of multiple parameters. In order to decrease total testing time, there is an interest in techniques that generate smaller test suites. In our previous work, we used graph techniques to produce high‐quality test suites. However, these techniques require a lot of computing power and memory, which is why this paper investigates distributed computing for t‐way testing. We first introduce our distributed graph colouring method, with new algorithms for building the graph and for colouring it. Second, we present our distributed hypergraph vertex covering method and a new heuristic. Third, we show how to build a distributed IPOG algorithm by leveraging either graph colouring or hypergraph vertex covering as vertical growth algorithms. Finally, we test these new methods on a computer cluster and compare them to existing t‐way testing tools. Combinatorial test generation is the process of generating sets of input parameters for a system under test, by considering interactions between t values of multiple parameters; the paper investigates the use of distributed algorithms to generate such test suites. It proposes reductions of the t‐way test suite generation to two problems on graphs and provides distributed algorithms to solve them; these algorithms are then used as vertical growth algorithms to build a distributed version of IPOG. Finally, we test these new methods on a computer cluster and compare them to existing t‐way testing tools.