Kotlin to Rust: The Hard Truths of Rewriting a High-Scale Data Plane
Momento engineer Ramya Krishnamoorthy lays out why they ditched their Kotlin/JVM data plane to wrestle tail-latency demons in their serverless caching platform. Garbage-collection pauses and unpredictable response times at scale drove them straight into Rust’s arms, where manual memory management and zero-cost abstractions deliver rock-solid, low-tail latencies.
Along the way, Ramya breaks down the real-world headaches of a full-blown rewrite—rewiring core abstractions, retraining teams, and tightening CI/CD pipelines—versus the gains: leaner runtimes, more consistent performance, and a production system that won’t be sidelined by a rogue GC cycle. For anyone curious if Rust lives up to its high-throughput hype, this candid deep-dive lays it all on the table.
Watch on YouTube
Top comments (0)