Windows performance modeling and data reduction using WatchTower and Argus

We describe and evaluate WatchTower, a system that simplifies the collection of Windows performance counter data, and Argus, a statistical methodology for evaluating this data and reducing the sheer volume of it. This is especially important when monitoring the performance of clusters for high performance computing. WatchTower’s overheads are comparable to those of Microsoft’s perfmon tool, while, unlike perfmon, it supports higher sample rates and is easily embedded into other software. Argus can reduce the behavior of a large number of raw performance counters into a few composite counters, or it can select a subset of the original counters that are statistically interesting. We are currently integrating WatchTower and Argus into a single system which we believe will be useful in a number of scenarios. Windows Performance Monitoring and Data Reduction using WatchTower and Argus Extended Abstract Michael W. Knop Praveen K. Paritosh Peter A. Dinda Jennifer M. Schopf knop, paritosh, pdinda, jms @cs.northwestern.edu Department of Computer Science Northwestern University 1890 Maple Avenue Evanston, IL 60201