Faculty Mentor: Caiwen Ding (UConn)

Description:

Infrastructure monitoring in data centers is often crucial in order to ensure that necessary system resources are consistently in good health and available for use. Maintaining this consistent monitoring can be time consuming and costly. To address this issue, this project details a study of (i) creating an effective and user-friendly metrics data monitoring service and (ii) using the data (CPU usage, memory usage, disk throughput, and network throughput) collected from the monitoring service to predict utility usage using machine learning. First, we collect customer infrastructure metrics data using the monitoring service.

Objectives:

We will then develop a machine learning algorithm to predict what customer computer metrics data will look like in the future based on previously detected trends. These predictions allow a customer to be alerted not only when infrastructure issues occur, but also when their systems may be vulnerable in the future. Through this project, students will gain knowledge of computer systems, data analytics, and machine learning.