Click here for full text:
Short term performance forecasting in enterprise systems
Powers, Rob; Goldszmidt, Moises; Cohen, Ira
Keyword(s): IT system performance; forecasting algorithms; time series analysis; Bayesian networks
Abstract: We use data mining and machine learning techniques to predict upcoming periods of high utilization or poor performance in enterprise systems. The objective is to automate assignment of resources to stabilize performance, (e.g., adding servers to a cluster) or opportunistic job scheduling (e.g., backups or virus scans). Two factors make this problem suitable for data mining techniques. First, there is abundant data given the state of current commercial monitoring and data collection tools for enterprise systems. Second, the complexity of these systems defies human characterization or static models. We formulate the problem as classification: given current and past information about the system's behavior, can we forecast whether the system will meet its performance targets over the next hour? Using real data gathered from several enterprise systems in Hewlett-Packard, we compare several approaches ranging from time series to Bayesian networks for classification. Besides establishing the predictive power of these approaches our study analyzes three dimensions that are important for their application as a stand alone tool: First, it quantifies the gain in accuracy of multivariate prediction methods over simple statistical univariate methods. Second, it quantifies the variations in accuracy when using different classes of system and workload features. This characterization is important for developing online resource allocation methods, where transfer functions from workload to system performance are desirable. Third, it establishes that models induced using combined data from various systems generalize well and are applicable to new systems, enabling accurate predictions on systems with insufficient data.
Back to Index