No abstract

In the present scenario, a server rack has multiple platforms attached to it, each designed to perform a different set of actions, thus, having different hardware requirements. To increase the throughput of such a platform either the hardware requirements are multiplied or the platform is replaced completely. This unoptimized method is rather expensive and inefficient.[1] This paper focuses on improving the performance of a system by providing accurate analysis and predict hardware requirements to improve overall throughput. For this, data logs are collected over a period of time which take performance data dumps of sensors connected to the platform via BMC. These sensors monitor the platform and measure its internal physical parameters. This data is then used to create a database and a training set. This set is used to train a machine learning algorithm which gives an efficient algorithm to analyze the present performance and give accurate prediction. This gives an optimal solution to increase throughput of a platform.[2] General TermsMachine learning, data logs, BMC, performance analysis and IPMI.