Bazel Remote Execution and Remote Caching
summary
Bazel remote caching and execution are implemented as a content‑addressable store, an action cache and an execution service tied together by action digests. A production rollout uses Hazelcast + Kubernetes for a REST cache backend, handles about 10,000 actions migrated so far, and reports a typical 10‑minute local build reduced to cache build 30s for a 10k‑step job. The open source BuildFarm implements CAS/action cache/execute, an LRU action cache, an operation queue and cancellation. REST backend lacks auth and chunking; gRPC is higher performance.
AI-written summary · about
chapters (12)
- 0:00Bazel remote execution
- 1:05Alpha’s build challenges
- 3:33Remote cache implementation
- 6:22Cache and execution work
- 8:51Benchmarking remote cache
- 10:56Remote cache best practices
- 12:20Remote cache limitations
- 14:05Remote execution and Buildfarm
- 17:40Buildfarm architecture
- 22:58Resource partitioning
- 25:52Buildfarm results
- 27:15Q&A
related talks
11mIntroducing Remote Bazel
01h 31mBazel
50mBuilding on Key – Keeping your Actions and Remote Executions in Tune + Q&A
20mAdopting Bazel in a quickly scaling organization
22mBazel
01h 01mIntroduction to Bazel
36m2 Hours to 10 Minutes: Building Rust Monorepos at Scale with Bazel RBE
24mFaster Builds With Remote Execution and Caching