Metric Cumulative Criterion

This page is accessed via Configuration and Administration > Metric Alert Management > Metric Cumulative Criterion.

Note: You must be a DBA to access this page.

A cumulative criterion is the use of one or two historical bases to trigger a cumulative threshold. For example, if you want to see if your bulk plans are taking longer to run now that they have historically, you would set up a cumulative criterion rule for comparing a recent value basis against a historical one. Then you would use that rule to define a threshold for the specific bulk plan performance metric. Note that cumulative criteria only trigger hourly, when they are recalculated based on the invocation of Persist Metrics. They are used to trigger performance trends as opposed to critical events. Cumulative criteria compares historical performance to either a fixed or historical value.

Missing Historical Data

The system captures metrics as it runs. However, if you are new to the product, you will not have any historical metrics to compare against. Alternatively, there could be gaps in the metrics if no data were captured as metrics are only persisted when non-zero. The system does not collect zeroes as data. For example, if logging did not occur, the database is not filled with zeroes to track the lack of logging. The following fields affect how your threshold handles missing data.

• Missing Value as Zeros: Select if you want to treat missing data as if they had been recorded as zeroes when comparing current transactions to ones in the past. E.g., assume we want to calculate the average daily bulk plan count over a week but only three days ran any bulkplans. Treating missing data as zeroes results in an average count of Bulk Plan Weekly Total Count/7. Otherwise, the average is calculated as Bulk Plan Weekly Total Count/3. Generally, treating missing data as zeroes is useful to avoid bias on count metrics; it's not needed for performance metrics.
• Minimum Comparison Value: Allows you to handle missing data by only triggering a threshold if the comparison basis exceeds the minimum value defined by this field. Only threshold values that exceed this minimum will trigger the threshold, regardless of the relative difference between the recent value and the comparison value. Only used for an historical comparison, where you compare past data to past data.
• Missing Comparison as Zeros: Select if you want to treat missing data as if they had be recorded as zeroes. If you want the missing data to be included as if zeroes were recorded, then select this field. Not selecting this field means that the missing data is not treated as a zero. Used when comparing past transactions to other past transactions.

Adding a Cumulative Criterion

1. Enter a Cumulative Criterion ID.
2. Select a Domain Name.
3. Enter an Historical Value. An historical basis to use as a value for comparison. Use this when comparing past metrics to ones that just occurred.
4. Select a Count Period. The Count Period of the criterion applies to Count Metrics (e.g. Total Logging) or Count Values (Count of Bulk Plan runs). It determines whether the historical basis (average or maximum) is based on hourly, daily, or weekly values. I.e., whether the threshold is based on hourly, daily, or weekly counts. The count period ensures we don't try to compare an hourly average/maximum count to a daily average/maximum count.
5. For the Missing Value as Zeros field, see Missing_Historical_Data.
6. Enter an Historical Multiplier. If there is a historical comparison, this is a multiplier against the historical value. For example, if the threshold fires when the maximum time over the last three days exceeds 300% of the two week historical average, the historical multiplier is 3.0. If there is no historical comparison, the multiplier acts as a fixed value for comparison. For example, if the threshold fires when the average time exceeds 2 hours, the multiplier is 2.0.
7. Enter an Historical Comparison. The historical basis to compare against the historical value. Use this when comparing past metrics to other past metrics.
8. For the Minimum Comparison Value. field, see Missing_Historical_Data.
9. For the Missing Comparison as Zeros field, see Missing_Historical_Data.
10. Click Finished.