How Go Helped Accelerate Software Delivery at the BBC
summary
BBC Digital Distribution handles 1,000 hours of catch‑up, 3–4 TB of live audio/video and about 13 million play events weekly across 17,000 GitHub repos, 4,000+ VM deployments, 2,000+ Lambdas and 600 AWS accounts. A deploy system is built on AWS CodePipeline/CodeBuild using pipeline definition files and Go CDK bindings, rewritten from a Docker-based tool into a Go binary in a day, plus Go Lambdas and dependency‑injected tests. It runs in production, exposes Grafana metrics, and cuts team tooling overhead that previously cost up to half a day per week.
AI-written summary · about
chapters (16)
- 0:00BBC background and role
- 1:00Engineering at BBC scale
- 2:37The problem with local tools
- 4:45Seeking standard solutions
- 6:11Adopting principles over tools
- 7:57Designing the new system
- 9:18Infrastructure as code
- 11:16Adopting Go for tools
- 12:58Benefits of Go in SRE
- 14:48Pipeline implementation details
- 18:50Extensibility and dogfooding
- 20:47Automation and visibility
- 21:49Testing Go lambdas
- 24:54Metrics and demonstration
- 25:47Lessons and future work
- 29:04Conclusion
