Optimizing Docker Builds - Successes, Failures, and Instrumented Observability

Explore strategies for optimizing Docker builds in this 20-minute conference talk from DockerCon 2021. Dive into successful techniques, common pitfalls, and the power of instrumented observability as speaker Nicole Rifkin shares insights on improving Docker deployment pipelines. Learn about layer caching, multi-stage builds, Buildkit inline caching, and the use of OpenTraces with Jaeger for enhanced visibility. Discover a simpler observability pipeline that offers flexibility with minimal overhead. Gain valuable knowledge on comparing builds, potential optimizations, and caching dependency installations. Conclude with key learnings and an opportunity for questions to further enhance your Docker build optimization skills.

Optimizing Docker
Deployment pipeline
Layer Caching Changing the order of execution can
Multi-stage builds Not only can multistage builds reduce parallellation and better caching
Our Docker build is still slow
Why do we care?
Buildkit inline caching
Changes without results
Opentraces with Jaeger
Insights from Jaeger
Monitoring tool overload
A simpler solution
Observability pipeline Simple, flexible, with very little overhead
Comparing builds
The fun questions
Optimizations we're exploring Caching dependency install
Learnings
Questions?