Performance projection of an application for large number of users involves predicting the maximum throughput that the application can achieve and the maximum number of users it can support. Factors affecting the maximum throughput can include both hardware and software resources of each of the servers associated with the application. In a multi-tier environment, the number of resources affecting the application performance can be quite large. Whenever any of these resources is bottlenecked, i.e., the resource utilization reaches close to 100%, the rate of increase of throughput drops. Further increase in the number of users beyond a certain point may result in reduction in throughput. For any enterprise application that intends to cater to a large number of users, it is desirable to know the maximum throughput it can achieve. This paper proposes a systematic technique for analyzing maximum throughput of any application that can be used irrespective of the test environment or production environment. Our technique computes maximum throughput with more than 95% accuracy in most scenarios. This technique can be useful in reducing the load testing effort and time.
S. Duttagupta and Nambiar, M., “Maximum throughput computation of an application in a multi-tier environment”, 2012 International Symposium on Performance Evaluation of Computer Telecommunication Systems (SPECTS). IEEE, Genoa, Italy, pp. 1-7, 2012.