Apply computational algorithms to ensure anomalous metric values are not a source of false positives

Traditional approaches to monitoring that employ static thresholds are now well acknowledged to be a significant source of alert noise. This is due in large part to the fact that monitoring data is inherently noisy with anomalous values regularly exceeding fixed threshold values even though no actual underlying problem or issue exists. More sophisticated methods of addressing this problem have emerged over the years that leverage statistical techniques to deal with an inherently noisy process.

NetSpyGlass was designed with an appreciation of the importance of an intelligent alerting system that is capable of leveraging a variety of computational algorithms to reduce the alert noise that leads to excessive false positive alerts. Because NetSpyGlass is a fully-programmable platform, users may implement custom computational algorithms or use any of those available with NetSpyGlass out of the box:

  • DBSCAN to recognize clustering and outlier characteristics in the values of monitoring data
  • Moving Average to recognize slowly trending characteristics in the values of monitoring data
  • Standard Deviation to recognize when the mean value of monitoring data moves outside preset range

With NetSpyGlass, there is no limit to the level of sophistication in computationally ensuring that inherently noisy monitoring data does not become a source of false positive alerts.

Leave a Reply

avatar
  Subscribe  
Notify of