FerroDruid — development record
FerroDruid is an Apache Druid–compatible OLAP engine in Rust — native queries and SQL against a Druid-shaped API, one binary instead of a JVM cluster. Live on AWS Marketplace.
Measured record (as of 2026-07-10)
| First commit | 2026-04-27 |
| Tags | v0.2.0 → v1.0.0 (2026-07-01) |
| Commits | 482 |
| Rust code lines (tokei) | 100,599 |
| Test annotations (tracked) | 2,390 |
| Latest activity | 2026-07-05 |
Quality gates passed
- Conformance harness that replays upstream-Druid query suites and compares responses against a real Druid — compatibility is asserted by diffing, not by reading docs.
- Covered by the nightly fuzz farm; findings feed the same reproduce → failing test → fix → green loop as every Ferro product.
- Same release gate as the rest of the fleet: adversarial review until two consecutive clean rounds.
Honest limitations
- Druid’s surface is large; conformance coverage is strongest on the query paths the harness replays. Ingestion-side parity (supervisors, exotic extensions) is narrower.
- Published benchmarks against a tuned Druid cluster do not exist yet — no performance multiplier is claimed here.