How do you monitor rolling deployments with APM?

The standard and recommended method for monitoring any deployment, including rolling deployments, in Datadog APM is to use the version tag. By instrumenting your application to add a version tag to its traces, you can distinguish between the old and new versions running concurrently during the rollout. This allows you to precisely compare key metrics like error rates, latency, and request throughput between versions on the APM Service Page. This comparison is critical for detecting regressions and ensuring the health of the new release as it is progressively deployed.

B. A rolling:true tag is not a standard Datadog convention and would not provide the necessary version-over-version comparison.

C. The source tag is used to identify the origin of telemetry (e.g., nginx), not for tracking application versions during deployments.

D. This is incorrect because option A describes the correct and standard procedure for monitoring deployments with Datadog APM.

1. Datadog Documentation, "Deployment Tracking". (docs.datadoghq.com/tracing/deploymenttracking/). The primary instruction is: "To enable Deployment Tracking for APM, add the version tag to your services' spans."

2. Datadog Documentation, "Getting Started with APM". (docs.datadoghq.com/tracing/gettingstarted/). The setup guides for various languages show how to configure the tracer with service, environment, and version tags.

3. Datadog Blog, "Monitor canary deployments with Datadog APM". (www.datadoghq.com/blog/monitor-canary-deployments-with-datadog-apm/). This article explicitly details using the version tag to compare performance between canary and primary versions.