Research On Clustering And Scheduling Of Software Testing Tasks Based On Cloud Computing Platform

With the development of cloud computing technology, more and more applications move tothe cloud computing platform, on one hand, because of cloud computing powerful resource capacityand reliability features,which can enhance these applications’ security and ensure them runefficiency; on the other hand, cloud computing can help more users take advantage of cheap,flexible services. The software testing is one kind of applications, because of its growing size andgrowing complexity, the traditional testing mode can not meet the growing demand, which needs tobe migrated to the cloud computing platform. Meanwhile, the testing tasks in the cloud computingplatform should be clustered and scheduled based on their properties. In this case, software testingcan be more efficiently expanded.The idea of this thesis is to move the software testing to the cloud computing platform, clusterthem according to the properties of the testing tasks, then make a reasonable scheduling based onthe resources provided by the cloud computing platform, which are to improve the efficiency oftesting tasks. The contribution of this thesis is moving the testing tasks to the cloud computingplatform. First, with improved K-means clustering algorithm, the testing tasks can be assigned todifferent clusters; second, the testing tasks on the cluster make up a lot of time and cause loadimbalance for scheduling with the FCFS algorithm and min-min algorithm, to solve these problems,an improved scheduling algorithm is put forward. Last, the thesis use cloudSim simulation tools forsimulation experiments. The experimental results show that: the improved clustering algorithm tosome extent reduces the time of the task clustering,thus improve the efficiency of clustering; on theother hand, the proposed scheduling policy, regardless of the total task completion has betterperformance on load balancing.