Automating Computer Bottleneck Detection with Belief Nets

We describe an application of belief networks to the diagnosis of bottlenecks in computer systems. The technique relies on a high­ level functional model of the interaction be­ tween application workloads, the Windows NT operating system, and system hardware. Given a workload description, the model pre­ dicts the values of observable system counters available from the Windows NT performance monitoring tool. Uncertainty in workloads, predictions, and counter values are charac­ terized with Gaussian distributions. During diagnostic inference, we use observed perfor­ mance monitor values to find the most prob­ able assignment to the workload parameters. In this paper we provide some background on automated bottleneck detection, describe the structure of the system model, and dis­ cuss empirical procedures for model calibra­ tion and verification. Part of the calibration process includes generating a dataset to es­ timate a multivariate Gaussian error model. Initial results in diagnosing bottlenecks are presented.