Enterprise management system and method which indicates chaotic behavior in system resource usage for more accurate modeling and prediction

Abstract

A set of algorithms for detecting a renewal power-tail behavior which often relates to chaotic system activities in one or more computer system resources of a distributed computing environment, i.e., an enterprise. Analysis and/or prediction software receives a set of metric data points from agent software on one or more computer systems. The analysis and/or prediction software performs three analytic tests relating to distinctive properties of power-tail distributions: a first test to determine whether the set of data points exhibits large deviations from the mean, a second test to determine whether the set of data points exhibits a high variance, and a third test to determine whether the set of the largest data points exhibits properties consistent with large values in a tail portion of a power-tail distribution. The tests can be performed in any order, and in other embodiments, fewer than three can be performed. Each test has two possible results: successful if the test indicates a likelihood of power-tail behavior, or unsuccessful if it indicates that power-tail behavior is unlikely. The results of the three tests are combined to determine the overall likelihood of a power-tail distribution. If all three tests are successful, then power-tail behavior is likely. If all three tests are unsuccessful, then power-tail behavior is unlikely. If the results are mixed, then typically more data or analysis is needed. The results are used for modeling and/or altering the configuration of the enterprise.