Fine-Grain Analysis (FGA): A Methodology for Analyzing Intermittent Performance Problems

Abstract

Maintaining adequate performance is essential for effectively managing a computer installation. This paper describes a methodology for analyzing intermittent performance problems - fine-grain analysis (FGA). FGA emphasizes the use of process-control charts (a type of time-series plot), both for detecting the presence of problems and for characterizing the underlying causes. We illustrate FGA by applying it to data collected from a computer system running IBM's operating system Virtual Machine System Product with the High Performance Option (VM/SP HPO).