Datadog CI/CD Optimization: The Case for Treating Your Pipeline as a Production Service
Enterprise Developer PlatformsMost engineering teams apply production-level observability to their deployed services and spreadsheet-level visibility to the pipelines that build them. Datadog is positioning CI/CD Optimization as the correction to that asymmetry.
What Happened
Datadog has released CI/CD Optimization, a platform capability that brings pipeline performance analytics, flaky test management, and test execution optimization into a unified view alongside existing Datadog observability data. It integrates across GitHub Actions, GitLab CI/CD, Jenkins, Azure DevOps, CircleCI, and Buildkite without requiring provider-specific instrumentation changes.
The core capabilities:
- Pipeline performance analysis: Tracks execution time, queue time, failure rates, and reliability trends across CI environments
- Flaky test detection: Surfaces tests with inconsistent pass/fail behavior and quantifies the CI time wasted on reruns
- Test impact analysis: Maps tests to code coverage and determines which tests are relevant to specific code changes, allowing irrelevant tests to be skipped on a given run
- Test parallelization: Optimizes test distribution across CI workers based on estimated test duration
Customer results from the release: Betterment reduced average build duration from approximately 40 minutes to under 10. The Browser Company reduced total CI pipeline time by 50%.
Why It Matters
The case Datadog makes is structural: CI pipelines are production infrastructure in all the ways that matter — they gate deployment velocity, accumulate technical debt in the form of unreliable tests, and consume real compute cost — but most teams don’t operate them with production-grade visibility. A 40-minute build with no instrumentation is a black box that costs engineering time and cloud compute on every commit, with no systematic way to diagnose why it takes 40 minutes instead of 10.
The flaky test problem is where the ROI becomes most concrete. Flaky tests — tests that pass and fail non-deterministically — generate reruns that consume CI minutes without providing signal. Teams that have quantified their flaky test cost typically find it represents 10–25% of CI compute spend and meaningfully larger proportions of developer wait time. Datadog’s existing pipeline monitoring work has been building toward this kind of holistic CI visibility for several release cycles.
Test impact analysis addresses a different inefficiency: running every test on every change regardless of what changed. For large test suites, this is a significant throughput constraint. Mapping tests to code coverage enables selective test execution — running the tests that are likely to catch regressions from a specific change, not the full suite for every commit.
The So What
CI/CD optimization has historically required custom tooling or dedicated platform engineering work — integrating coverage data, building flakiness dashboards, implementing selective test execution. Datadog’s platform integration means that teams already running Datadog for production observability can extend the same visibility model into their pipelines without building bespoke tooling. For organizations where agentic CI/CD workflows are increasing pipeline complexity, having unified telemetry across the pipeline and the services it deploys to becomes particularly valuable — the same trace context that spans production requests can now span the build jobs that deployed them.
Source: Datadog
Content created with AI assistance and reviewed for accuracy.
Join the conversation
Stack Insiders is our free community for readers who want to go deeper — share resources, ask questions, and connect with others across every vertical we cover.
Join Stack Insiders →