Case Study: HUD achieves real-time production code visibility with ClickHouse Cloud

How Hud is building the first runtime code sensor with ClickHouse Cloud

Hud built the “world’s first runtime code sensor” to observe how code behaves in production at both the function and business level, but their vision required processing huge volumes of function-level telemetry with very low overhead. As their product scaled, they needed a database that could handle high-cardinality telemetry, real-time correlation with metadata, and version-aware analysis across continuous deployments. Hud turned to ClickHouse Cloud to replace an initial Postgres-based prototype and support this demanding observability workload.

Using ClickHouse, Hud redesigned its ingestion pipeline to run enrichment directly inside the database with Null engine tables, dictionaries, and ReplacingMergeTree tables, eliminating Redis and reducing network and memory overhead. The result was a simpler, faster, and more stable system that now ingests hundreds of MB/s of raw JSON, compresses that to tens of MB/s in storage, and stores more than 11 TB of telemetry in ClickHouse Cloud. With ClickHouse, Hud can detect degradations within minutes of a release and surface root-cause context directly to engineers in the IDE.

HUD

May Walter

Co-Founder and CTO