Today we’re launching APM dashboard templates for Datadog, so platform and SRE teams can get best practices observability into GraphOS Router performance in just minutes. Previously teams would need to determine the important information to monitor, what telemetry is available, and then how to configure the dashboards. This could take hours or days of an engineers time to initially set up and refine over time.
Whether you’re just turning on tracing or you’ve been running Apollo GraphOS at scale for years, these dashboards provide a clear, shared view of the GraphOS Router, supergraph, and subgraph health in minutes, just import and go.
Turn-key observability for Datadog
Operating a supergraph is a team sport. When latency spikes or error rates climb, you need to answer:
- Is it the router, a specific subgraph, or client behavior?
- Is the issue limited to a region, version, or deployment?
- What changed right before the incident?
Our Datadog templates + GraphOS Router instrumentation answer those questions:
- Clear, consistent operations: Spans and metrics are mapped into Datadog with stable operation and resource names, so charts and traces line up.
- First-class GraphQL error tracking: GraphQL errors are promoted into APM error views and correlated with latency.
- Supergraph ↔ subgraph drill-down: See end-to-end behavior, then pivot to the slow or noisy subgraph instantly.
What you’ll see in Datadog
- Golden signals at the top: Requests, p95/p99, error rate, saturation.
- GraphQL errors: Surfaced as first-class errors with example traces.
- Subgraph focus views: Rank subgraphs by latency contribution and error impact.
- Operation insights: Identify the slowest or flakiest operations and track regressions by version.
- Deploy & version overlays: Correlate performance shifts with version rollouts and activity.
Get started with the new dashboard templates
Connect Apollo Router to Datadog
Set up metric ingestion for Datadog based on your preferred method:
- Datadog OpenTelemetry Collector (DDOT) – straightforward if you’re already Datadog-centric.
- OpenTelemetry Collector – vendor-neutral with powerful processors/sampling.
- Datadog Agent – direct OTLP ingestion alongside your existing Datadog infra agent.
- Agentless – when you want no agents at all.
Configure the Router to emit the needed instrumentation
Add the required opentelemetry settings to the Router’s configuration file, tailoring resource.name and tags to your conventions (service, environment, version, region) if needed. Restart the router so that the new configuration can be applied.
Install the dashboard template
Inside the Datadog console, import the new dashboards: Dashboards → New Dashboard → Import JSON and then paste our template.
Set the template variables (service/env/version/region) to match your Datadog tags. You should see the dashboard graphs begin to populate within minutes.
That’s it! Your supergraph’s APM story is live in Datadog
Managing your observability costs
In order to optimize observability costs, we recommend keeping resource.name low cardinality by using stable values such as operation_kind or subgraph.name, and avoid embedding IDs or other highly variable strings.
For high-throughput services, enable tail-based sampling in your OTLP collector to retain the most useful traces while controlling cost, and explicitly exempt error traces so failures always surface.
Finally, apply Datadog’s Unified Service Tagging (service, env, version) consistently across traces and metrics so dashboard filters and version overlays behave reliably across teams.
Questions or feedback?
Check out our documentation to learn more about setting up and using these dashboards. We’d love to hear how these dashboards are working for your team and what you want to see next.
