OpenTofu's Performance Pillars, Pitfalls, and Profiling
Talkat OpenTofu Day North America 2025 · on Monday, 2025-11-10· at 13:35· Building B | Level 2 | B203·
summary
OpenTofu adds OpenTelemetry traces and a Go CPU profiler to pinpoint where time and memory go; profiling shows a contrived config with 1,000 duplicated resources produces ~2,000 plan operations, the run takes ~10s versus an ideal ~2s, and GC plus repeated decoding dominates. Changes remove deep state copies, reduce lock contention, avoid syncing state 2,000 times and speed config loading, cutting a plan+apply from ~42s to ~10s and a 2m apply down to ~27s. A new architecture and granular state storage are planned.
AI-written summary · about
chapters (10)
- 0:00OpenTofu introduction
- 1:07Why performance matters
- 2:53Core components and factors
- 5:02OpenTelemetry diagnostics
- 8:32CPU profiling bottlenecks
- 11:02Large-scale plan analysis
- 14:10OpenTofu 1.8 improvements
- 16:38Future optimizations
- 18:24User performance help
- 20:15Q&A
related talks
44mGo Profiling and Observability from Scratch
40mTwo Go programs, three different profiling techniques, all in 40 minutes
19mInstrumenting Go Apps With OpenTelemetry
30mMastering Distributed Tracing with Go
34mOpenTelemetry for Dummies: Instrumenting Go Apps
40mProfiling real-world golang service
39mEfficient Debugging and Logging with OpenTelemetry in Go
51mObscure Go Optimisations