Metrics

Describes metric reporting capabilities provided by Philter.

Philter collects metrics while running to provide insights into its operation and the text being processed. The metrics collected include a count of the documents processed by Philter, counts of the types of sensitive information identified per type, and the entity confidence values of entities extracted by non-deterministic natural language processing methods. These metrics, by default, are reported to standard out, but can also be reported via JMX, Amazon CloudWatch, and Datadog.

Philter supports reporting metrics via JMX, Amazon CloudWatch, and Datadog.

Reporting Metrics to Amazon CloudWatch

To enable Philter metric reporting to Amazon CloudWatch modify Philter's application settings to set the AWS properties as detailed in the Settings.

Metrics will be published to CloudWatch every 60 seconds when enabled.

The AWS IAM user or role being used should have PutMetricData permissions:

{
"Version": "2012-10-17",
"Statement": [
{
"Sid": "VisualEditor0",
"Effect": "Allow",
"Action": [
"cloudwatch:PutMetricData"
],
"Resource": "*"
}
]
}

The metrics will be published to the Amazon CloudWatch namespace provided in Philter's settings. Amazon CloudWatch can then be used to visualize the metrics, set performance alarms, or perform other integrations with AWS services.

Philter metrics reported and visualized in Amazon CloudWatch.

Reporting Metrics to Datadog

Metrics will be published to Datadog every 60 seconds when enabled.

Metrics published to Datadog will have a philter prefix.

Philter metrics in Datadog's Metrics Summary.

The metrics can be used to make graphs and dashboards.

Example Datadog graphs of select Philter metrics.

Reporting Metrics to JMX

Metrics in JMX can be viewed using visualvm or similar tool.

Metrics Collected and Reported

The listing below shows an example of the metrics Philter collects and writes to standard out while running. The metrics reported to supported services such as JMX, Amazon CloudWatch and Datadog will contain the same metrics but may be represented or visualized differently between the services.

The metrics collected include:

  • A cumulative count of each type of sensitive information across all contexts and documents.

  • The total count of documents processed.

These metrics will be reset when Philter is stopped and restarted.