The Software Factory
summary
Large orgs with 1,000+ developers and millions of lines of code get predictable: hundreds of bespoke build scripts, flaky integration branches, and queues on CI. The solution presented uses pull requests as contracts, Jenkins driving a conveyor of builds, Kubernetes for elastic build/test capacity (replacing 50 Jenkins masters and ~400 static slaves), GitHub/GitOps for traceability, and a Kafka firehose into Postgres for dashboards. Builds from developer to integration take up to eight hours and full promotion can span two weeks, and the pipeline and visibility stack are already reducing firefighting.
AI-written summary · about
chapters (17)
- 0:00Introduction
- 2:30Problems at Scale
- 3:25Back to Basics
- 6:05Tools
- 10:23Customer
- 14:55Developers
- 16:26Containers
- 17:18Arrow
- 17:29Getops
- 19:41Common Processes
- 21:30Visibility
- 27:57Visibility Layer
- 30:15DevOps Layer
- 31:53Any Questions
- 32:52What happens to D
- 38:18Quality bar and frequency
- 42:43Kafka topics
related talks
43mDevOps for Mere Mortals
29mFrom scheduled builds to commit triggered pipelines
27mHow We Streamlined Our SDLC with Observability - from GitHUB via Jenkins, Harbor, Argo to K8S
22mA Healthier Software Development Life Cycle with DevOps
01h 01mA team's transition to Continuous Delivery
27mConcept to Code to Runtime - Simplify, Enable, Exploit
50mWhy You Need a Software Delivery Machine
54mFrom Code to Customer