Determining an appropriate threshold is essential while monitoring business-critical applications. Ops Center Analyzer analyzes the peak, normal, and low volume phases based on the historical data and adjusts the monitoring thresholds accordingly. Automating the threshold calculation eliminates false alerts and reduces the number of alerts to investigate, which might otherwise become a management overhead.
The application workloads might vary at different times of the day or week. For example, the workload pattern of an OLTP application might be different on weekdays and weekends. You can manage varying workloads that occur at different time periods for an application by creating monitoring plans. The system analyzes the performance data accumulated in the scheduled baseline period for computing the dynamic threshold values.
The following example shows the response time metrics of a business-critical application monitored over time and how the system derives the automatic threshold values based on the past performance. The high-level steps the system uses to calculate the automatic baseline values are as follows:
- Analyzes historical data for identifying the performance patterns in the specified baseline period.
- Detects and removes the occasional outliers: In the following example, the data points that deviate from the norm represent the outliers. The system ignores the outliers appearing at irregular intervals to calculate an appropriate threshold value.
- Calculates the maximum value: The upper limit of the values in the normal range is used to calculate the maximum value. After determining the maximum value, the system adds the margin of error to the computed value.
- Determines the weighted average: The weighted average derives the threshold values based on the past performance trends over a specified time period.