Communication-efficient distributed monitoring of thresholded counts
AssigneeAlcatel-Lucent USA Inc.
InventorsGraham R. Cormode, Ram Keralapura, Jeyashankher Sr
Life of this patent
Estimating the term…
Abstract
A system, method, and computer program product for distributed monitoring of local thresholds at each of a number of monitoring nodes and initiating communication only after the locally observed data exceeds the local threshold. Both static thresholds and adaptive thresholds are considered. In the static case, a combination of two alternate strategies for considering thresholds minimizes communication overhead. In the adaptive case, local thresholds are adjusted based on the observed distributions of updated information in the distributed monitoring system. Both approaches yield significant savings over the naïve approach of performing processing at a centralized location.
Claims17
A method for distributed network monitoring, comprising:
- Assigning, by a computer, at least one fault condition threshold to each of a plurality of sites, each fault condition threshold assignment is blended so that the fault condition threshold assignment is a linear combination of a uniform threshold and a proportional threshold, the fault condition thresholds being determined using each of a uniform fault condition threshold determination method and a proportional fault condition threshold determination method;
- applying at least one determined fault condition threshold in a manner tending to minimize a number of fault condition updates generated by a corresponding site;
- receiving at least one update from each a site only after the a fault condition threshold is violated at the site;
- and
- estimating a total count of violations for all the sites based on the updates received.
A system for distributed network monitoring, comprising:
- a plurality of sites each having at least one fault condition threshold, the fault condition thresholds being determined using each of a uniform fault condition threshold determination method and a proportional fault condition threshold determination method each site only sending an update after the fault condition threshold is violated;
- and
- a coordinator for receiving the updates, assigning the at least one fault condition threshold, each fault condition threshold assignment is blended so that the fault condition threshold assignment is a linear combination of a uniform threshold and a proportional threshold and the coordinator estimates a total count of violations for all the sites based on the updates received.
A non-transitory computer-readable medium storing instructions, which when executed by a processor direct an operation of a computer to perform a method for distributed network monitoring, the method comprising:
- assigning at least one fault condition threshold to each site of a plurality of sites, each fault condition threshold assignment is blended so that the fault condition threshold assignment is a linear combination of a uniform threshold and a proportional threshold, the fault condition thresholds being determined using each of a uniform fault condition threshold determination method and a proportional fault condition threshold determination method;
- applying at least one determined fault condition threshold in a manner tending to minimize a number of fault condition updates generated by a corresponding site;
- receiving at least one update from a site only after a fault condition threshold is violated at the site;
- and
- estimating a total count of violations for all the sites based on the updates received.
Dependencies are inferred from the claim text.