← All work
A Rust ingestion service for high-volume telemetry
- Client
- B2B observability startup
- Year
- 2024
- Stack
- rust, tokio, observability, performance
Sustained 12x the previous peak throughput on the same hardware, with zero dropped events.
The existing ingestion service, written in Node.js, was dropping events under peak load and had no clear backpressure strategy. I rebuilt the hot path in Rust using Tokio, added a proper backpressure and batching strategy against the downstream store, and kept the surrounding TypeScript control plane untouched so the client's team could keep shipping features against it.