Proposed Framework for Performance Tuning in Enterprise Applications using Machine Learning

IT systems and enterprise applications are crucial and their failure can lead to production losses, data loss, and missed commercial opportunities, resulting in costly efforts to recover. Nowadays, most of the applications are data-intensive, as a result of which the systems and the applications are becoming more and more complex in terms of their configurations. There are hundreds of configuration knobs that affect the performance of the system. Due to these changing environments and requirements of the software systems, the systems should be robust, versatile, autonomous, energy efficient and customizable at run-time. These requirements of the system to maintain its performance has surpassed human capabilities. Therefore, this research study proposes a novel approach to automatically tune the performance of the system using machine learning approaches. Here, the idea is to propose a technique that will identify anomalous behavior of the system because of one or more configurational parameters and auto-remediate the system to its normal state.