Use one of the logs templates to create a log pipeline.
- Archive Logs
- Dual Ship Logs
- Generate Log-based Metrics
- Log Enrichment
- Log Volume Control
- Sensitive Data Redaction
- Split Logs
See Set Up Pipelines for more information on setting up a source, processors, and destinations.
Use the Metric Tag Governance template to create a metrics pipeline.
See Set Up Pipelines for more information on setting up a source, processors and destination.
Metrics data
Metrics sent to Observability Pipelines include the following:
name: The metric name.kind: There are two kinds of metrics:absolute metrics: Represents the current value of a measurement at the time it is reported.incremental metrics: Represents the change in a measurement since the last reported value, which the system aggregates over time.
value: The metric type:countergaugedistributionhistogram
timestamp: The date and time the metric is created.tags: Includes tags such as host.
Whether a received metric is incremental or absolute depends on the source. For example, metrics from OpenTelemetry can either be incremental or absolute based on their temporality. The following table is an example of an OTel counter metric sent with delta versus cumulative temporality.
| Metric Type | Incremental | Absolute |
|---|
| Counter | Sent as deltas: +2, +4, +6 | Sent as cumulative sum: 2, 6, 10 |
An example of a metric:
{
"name":"datadog.agent.retry_queue_duration.bytes_per_sec",
"tags":{
"agent":"core",
"domain":"https://7-72-3-app.agent.datadoghq.com",
"host":"COMP-YGVQDJG75L",
"source_type_name":"System",
"env:prod"
},
"timestamp":"2025-11-28T13:03:09Z",
"kind":"absolute",
"gauge":{"value":454.1372767857143}
}
Metric types
The available metric types:
| Metric type | Description | Example |
|---|
| COUNTER | The total number of event occurrences in one time interval. Can be reset to zero, but cannot be decreased. | You want to count the number of logs with status:error. |
| GAUGE | A snapshot of a value at the time it is reported. | You want to track the latest CPU utilization for each host. |
| HISTOGRAM | Statistical aggregations (avg, min, max, count, median, percentiles) computed per host by the Datadog Agent in one time interval, then sent to Datadog. | You want per-host request latency aggregations from each web server. |
| DISTRIBUTION | Raw values sent to Datadog so percentile aggregations are computed server-side, globally across every host reporting the metric in one time interval. | You want the global p95 latency of an API endpoint, calculated across every host serving it. |
See Metric Types for more information.
