Building a Distributed Build System at Google Scale
summary
Google-scale means 30,000 engineers, ~45,000 commits per day, a ~2 billion‑line monorepo and BuildRabbit handling 5 million build and test requests per day producing petabytes of artifacts. The system layers Basil (Google's build tool) under a distributed service that migrates from a client push model to a persistent queue pull model, splits artifact and progress storage, and proves backends first. The rearchitecture swaps execution engines across five teams in three zones with a zero downtime engine swap and months‑long compatibility window.
AI-written summary · about
chapters (4)
related talks
30mBuilding Stable, Scalable & Agile Distributed Systems at Enterprise Scale
30mBuilding distributed systems is hard - case study (...)
11mRemote execution with 100 to 100,000 CPUs
29mUnlocking Bazel's Potential with Monorepo: A Key to Efficient Software Development
52mHow to build a planet-scale data replication...
29mHow to Replace a Jet Engine of your System In Flight
47mWeb-Scale IT Engineering.
43mScalability Intro – Shyam Jeedigunta, Google & Bob Wise, AWS (Any Skill Level)