A percentile is a value below which a given percentage of measurements in a set falls. For example, a 95th percentile value of 150 ms means that 95% of all values are 150 ms or less. This page describes how to modify default percentile collection. The Metric Browser describes what percentile metrics are and how to interpret them.
Disable Percentile Metric Collection
The Java Agent and the .NET Agent capture percentile metrics by default. Disable percentile metric collection in the Configure Percentile Metrics panel on the Configuration > Slow Transaction Thresholds window. Alternatively, manually set the disable-percentile-metrics node property to 'True' to prevent the agent from collecting percentile metrics.
Specify Percentiles to Collect
By default, app agents capture the 95th percentile. You can indicate 5 whole numbers between 1 and 99 as Percentiles to Collect on the Configuration > Slow Transaction Thresholds window. You can apply the configuration changes to all existing business transactions or only to new transactions discovered after the configuration change.
Modify the Percentile Metric Algorithm
The agent uses one of the following methods to calculate percentile metrics:
- P Square algorithm (default): This option consumes the least amount of storage and incurs the least amount of CPU overhead. The accuracy of the percentile calculated varies depending on the nature of the distribution of the response times. You should use this option unless you doubt the accuracy of the percentiles presented.
- Quantile Digest algorithm: This option consumes slightly more storage and CPU overhead for the machine where the agent is running, but may offer better percentiles depending on how the response times are distributed.
To change the algorithm the agent uses to calculate percentiles set the percentile-method-option app agent node property.