Automatic performance analysis of cloud based load testing of web-application & its comparison with traditional load testing

Abstract

Recently many application are migrated towards cloud computing. It becomes an emerging field of research in software testing; moreover it is non-trivial to evaluate the performance of cloud services. Performance and load testing are one of the dominant means to evaluate the web-application performance. At the end of load testing performance analyst have to analyze thousands of performance counters in both scenarios traditional as well as in cloud based load testing. These performance counters consist of run time system and web application properties such as resource consumption, response time, Memory utilization, throughput, Disk input output, latency, network traffic, delay. Performance analyst analyzes these performance measures manually, find out if the application meets service level agreement or not. It is very time consuming and error prone method, so to resolve this issue in this paper we proposed an approach to detect the performance deviation in cloud based load testing compare with the traditional load testing. It also helps performance analyst to compare load test more efficiently in order to detect performance deviation moreover it provides manageable set of important program counter to analyst for further and efficient root cause analysis. This approach is verified on the data obtain by performing load testing of web-application using J-meter in case of traditional and blaze meter in cloud based load testing. Our proposed approach provide up to 90% of reduction in the set of performance counter and 96% precision while detecting performance deviation with few false positives.