System stability prediction using prolonged burst detection of time series data

摘要

Embodiments of the invention relate to a system comprising a processor, a burst detection module executing on the processor, and a resource monitor. The burst detection module is configured to receive a set of resource usages samples measuring an availability of a resource, calculate an absolute moving average (AMA) of the set of resource usage samples, calculate a mean dispersion of the set of resource usage samples, and determine that the set of resource usage samples comprises an aberrant sample by comparing the AMA to the mean dispersion. The resource monitor is configured to execute a recovery procedure in response to the determination that the set of resource usage samples comprise the aberrant sample.

主权项

1. A system comprising:
a processor; a burst detection module executing on the processor configured to:
receive a set of resource usages samples measuring an availability of a resource;calculate an absolute moving average (AMA) of the set of resource usage samples;calculate a mean dispersion of the set of resource usage samples; anddetermine that the set of resource usage samples comprises an aberrant sample by comparing the AMA to the mean dispersion; and a resource monitor configured to:
execute a recovery procedure in response to the determination that the set of resource usage samples comprise the aberrant sample.