SOFTWARE PERFORMANCE PREDICTION USING LINEAR REGRESSION

Abstract – Performance prediction is required for succesfull management of software levels. It is of crucial importance especially in large distributed systems. Timely and proper planning allows for reduction of operation costs and increase of service levels. One of the ways to perform prediction is to employ trend analysis. In this paper linear regression is employed in order to predict performance of a software application, based on data obtained using continuous monitoring. DProf system, based on Kieker framework was used for data gathering