Arun Kumar Dhakshinamoorthy

Topic: Adaptive Learner

Abstract: The process of performance testing has improved many times over the course of its development as an industry. However, the automation of scenario design and performance test execution is still to be developed. Due to the increasing number of testing requirements for new and old applications it is becoming hard for the performance tester to design a scenario and the optimise it based on the server metrices obtained following testing. Tester must manually adjust the load based on the resource utilization and tune the performance characteristics of the server and the application settings manually. This requires several tests runs of the different test scenarios which required time and effort for the testing resources.

The Objective of the paper is to develop a solution to automate the scenario design process and performance tuning. A Python based machine learning application called the “Learner” will be part of the continuous integration model and will learn the behaviour of the application residing in the server, it will then design a pattern of load that needs to be generated by the performance testing tools. This tool controls the load generators by providing an adaptive feedback from the APM Tools.

The Learner will have a default pattern of testing which is then altered based on the feedback received from the server metrics. The Learner will also run sequence of tests by optimizing the server characteristics to get the best SLA and NFT achieved with the given set of resources for testing.

The Key benefits of the Paper
The final report generated by the performance testing tools will provide the optimum load that a server will be able to handle in all the conditions with optimum tuning of server characteristics.

Introducing the “Learner” in continuous integration and the testing lifecycle process will have an immense improvement on agility and speed of delivery. Thus, increasing the productivity and cost avoidance to the organization.