How can I reduce build times in my CI/CD pipeline when using devops best practices?

All answers to this question.

Reducing build latency is a core pillar of a mature <devops> culture. You should first analyze your Dockerfile structure; ensuring that the least frequently changed layers are at the top will maximize cache hits. Beyond that, moving to ephemeral Kubernetes agents for your Jenkins builds allows for parallel processing which can drastically cut down the time spent in the queue. We implemented a shared persistent volume for the package manager's cache (like npm or maven), and it shaved 15 minutes off our total build time. Also, consider "Shallow Clones" in Git to avoid downloading the entire history if you only need the latest commit for the build process.

   Answered 2025-03-16 by Carol Martinez

Are you currently monitoring your resource utilization during the build, or are you just assuming the <devops> pipeline configuration is the main culprit for the delay?

   Answered 2025-03-18 by Jeffrey Hall

• Jeffrey, we did a quick check on the node metrics and realized we are hitting CPU throttles during the unit testing phase. Do you think increasing the instance size is a sustainable fix, or should we be looking into test splitting to run them across multiple smaller containers instead? I'm worried about the cost implications of just throwing bigger hardware at the problem without fixing the underlying efficiency.

     Commented 2025-03-19 by Raymond Scott

In our <devops> workflow, we switched to Kaniko for building container images inside Kubernetes. It handles caching much better in a daemon-less environment.

   Answered 2025-03-21 by Brenda Allen

• I agree with Brenda. Kaniko was a lifesaver for our security team too, since it doesn't require privileged access to the Docker daemon.

     Commented 2025-03-22 by Gregory Nelson