Datadog Integration

MetricsHub integrates seamlessly with your Datadog environment. The MetricsHub app, available through the Datadog Marketplace, includes a collection of dashboards and monitors designed to collect and expose observability and sustainability data for your IT infrastructure in a turn-key solution.

Integrating MetricsHub with your Datadog SaaS platform only requires a few installation and configuration steps.

IMPORTANT
To ensure seamless integration of MetricsHub with Datadog, all metrics collected by MetricsHub must be prefixed with metricshub., as detailed below.

Although this documentation uses standard OpenTelemetry metric names (e.g., system.cpu.time), these metrics will appear in Datadog under the metricshub. namespace (e.g., metricshub.system.cpu.time).

The dashboards and monitors provided with the MetricsHub integration on the Datadog Marketplace are designed to query metrics using this prefix.

Prerequisites

Before you can start viewing the metrics collected by MetricsHub in Datadog, you must have:

1. Subscribed to MetricsHub from the Datadog Marketplace
2. Created an API key in Datadog as explained in the Datadog User Documentation
3. Installed MetricsHub on one or more systems that has network access to the physical servers, switches and storage systems you need to monitor
4. Configured resource monitoring.

Configuring the integration

Pushing metrics to Datadog

1. Edit the exporters section of the otel/otel-config.yaml configuration file:

   exporters:
     datadog/api:
       api:
         key: <apikey>
         # site: datadoghq.com
         # site: datadoghq.eu
         # site: ddog-gov.com
         # For further details, refer to https://docs.datadoghq.com/getting_started/site/
       metrics:
         resource_attributes_as_tags: true

   where <apikey> corresponds to your Datadog API key.

2. To ensure dashboards and monitors provided with the MetricsHub integration function correctly, add a transform processor to prefix the metrics collected by MetricsHub with metricshub.:

   transform/datadog:
     # Apply transformations to all metrics
     metric_statements:
       - context: datapoint
         statements:
           # prefix all metrics by `metricshub.` (except `otelcol` metrics, which come from the OpenTelemetry Collector itself)
           - set(metric.name, Concat(["metricshub", metric.name], ".")) where not IsMatch(metric.name, "^(metricshub\\.|otelcol_)")

3. Declare the receiver, processor, and exporter in a separate pipeline dedicated to Datadog, to avoid affecting other exporters with metrics prefixed with metricshub.:

   service:
     extensions: [health_check, basicauth, basicauth/partner]
     pipelines:
       # Dedicated pipeline for Datadog
       metrics/datadog:
         receivers: [otlp, prometheus/internal]
         processors: [memory_limiter, batch, transform/datadog]
         exporters: [datadog/api]

4. Restart MetricsHub to apply your changes.

Refer to Sending Telemetry to Observability Platforms for more details.

Configuring sustainability settings

To ensure dashboards are properly populated, you must configure the sustainability metrics hw.site.carbon_intensity, hw.site.electricity_cost, and hw.site.pue as explained in Configure Sustainability Metrics.

Adding monitors

To be notified in Datadog about any hardware failure, go to Monitors > New Monitor and add all the Recommended monitors for MetricsHub.

Creating your own monitors based on the ones listed as recommended allows you to customize the notification settings of each monitor.

Using the MetricsHub dashboards

MetricsHub comes with the following dashboards which leverage the metrics collected by MetricsHub:

Dashboard	Description
MetricsHub Hardware Overview	Overview of all monitored hosts, with a focus on sustainability
MetricsHub Hardware Site	Hardware metrics associated to one site (a data center or a server room) and its monitored hosts
MetricsHub Hardware Host	Hardware metrics and alerts associated to one host and its internal devices
MetricsHub System Performance	System performance and capacity metrics for Linux and Windows hosts