Continuous Delivery with Containers: The Good, Bad, & Ugly
summary
Containers become the single binary pushed through CI/CD and must be tested, labeled and versioned; Docker images + Kubernetes orchestration change runtime behavior (JVMs that aren’t cgroup aware see 32 cores and mis-tune GC, containers exhaust entropy and block crypto ops). Practical fixes: use semantic versioning, add labels/metadata or external registries, run scanners like Clair (scans returned dozens of CVEs), prefer a sidecar for test tooling, try Telepresence for local debug, and explore Grafeas/Kritis for signed attestations.
AI-written summary · about
chapters (11)
- 0:00Introduction to containers
- 5:19Continuous delivery overview
- 9:11Containers impact on CD
- 13:44Developer experience with containers
- 20:11Dockerfile and configuration
- 25:07Metadata and versioning
- 31:46Testing and security
- 39:04Security and scalability trade-offs
- 41:11Runtime nuances and pitfalls
- 43:58Observability and tooling
- 45:00Summary and key takeaways
related talks
40mContinuous Delivery with Containers: The Good, the Bad, and the Ugly
52mContinuous Delivery with Containers: The Good, the Bad, and the Ugly
56mDevOps in a containerized world
26mContainerization crossroads: Experiences-Experiments-Expectations
44mContain yourselves
49mContinuous Delivery at Docker age
26mCI/CD with Kubernetes - the Good, the Bad and the Ugly
37mDesigning Cloud Ready Applications