We rewrote a Go service in Rust and our velocity tanked for a quarter.

Short summary

A startup rewrote a Go backend in Rust and lost a full quarter of feature velocity despite real performance gains. Citing similar experiences from Discord, Vercel, and InfluxData, the author argues that language migrations are business decisions disguised as engineering ones. The lesson: only do a full rewrite if you can afford to build a quarter of what you already have, otherwise strangle the old service incrementally.

• •Rewriting Go to Rust delivered 12x faster API responses but cut sprint velocity by 65% for a quarter
• •Discord, Vercel, and InfluxData all faced multi-year logistical costs from full rewrites
• •Treat language migration as a business decision; do incremental rewrites unless you have time and money to spare

Generated with AI, which can make mistakes.