System performance analysis is the process of gaining a deeper understanding of those aspects of a computing system that affect its ability to perform its many functions as efficiently as possible. This is a complex undertaking requiring a great deal of experience, expertise and specialized knowledge of the system - its hardware, software and environment - in order to understand and then take action to improve performance.

This talk explores recent research in automated system performance analysis using a dynamic ensemble approach, statistical methods, causal inference, anomaly detection and explainable AI techniques. This new approach allows a collaborating human analyst to rapidly distill vast amounts of data, gaining understanding and insight into the major contributors to performance in real time, leading to improved system understanding for both optimization and root cause analysis tasks.