A multi agents based approach for testing MapReduce systems

MapReduce has emerged over the past few years as the most popular framework to build large-scale data based applications. However, the execution of such applications on large clusters produces several failures due to bugs or hardware problems. Testing such issues becomes very complex especially if the implementation under test (IUT) has to respect timing constraints. In this paper, we describe through two algorithms the test procedure in the side of the Mapper and the Reducer. Afterwards, we propose the use of Multi-Agents systems to model and simulate our approach for testing the MapReduce based implementation. The new proposed architecture will operate in an environment with timing constraints. Thereby taking into consideration the temporal properties in the specification of the behavior of conformance testing, it will provide a higher degree of expressiveness and flexibility. Finally, the simulation of our prototype is under experimentation in a MadKit Environment.