Performance Baselines

Changes in use levels, server memory leaks, and other abnormalities cause service-impacting events. Yet, it is easy to miss subtle changes in behavior when you rely on alerts based on predetermined, static thresholds only. These subtle changes often foreshadow a potential service disruption, if detected.

IBM SevOne Network Performance Management (NPM) uses machine learning to establish a baseline for every performance metric it collects. This provides you with an understanding what is "normal" behavior for any given time of day and day of the week. Alerts based on standard deviation from baseline performance notify you when exception conditions occur.

This method provides a more reliable predictor of service-impacting events. It also decreases the number of false-positive alerts.

For example, if your company always runs a backup procedure at 3:00 a.m., you don't want a daily alert about high bandwidth usage. But you would want an alert when an unexpected spike occurs during working hours due to a unique user-initiated action.

How you benefit:

  • Instant alerts when performance metrics deviate from expected behavior
  • Clear visualizations that compare typical performance of your infrastructure to real-time metrics
  • Better data when you don't know acceptable performance ranges for certain monitored objects
  • Greater granularity with baselines calculated at 15 minute intervals, not hourly intervals
  • Fewer false-positive alerts caused by static threshold violations that aren't noteworthy