Case Study: LangChain scales LangSmith observability and analytics with ClickHouse

Why we Choose ClickHouse to Power LangSmith

LangChain, the team behind LangSmith, needed a way to power observability and evaluation for LLM applications as usage grew from lightweight logging to high-volume production workloads. Their Postgres-only setup struggled with ingestion volume, analytical queries, and lock contention, especially for trace exploration, monitoring charts, and filtering data in real time. ClickHouse Cloud became the analytical database behind LangSmith’s hosted offering.

With ClickHouse, LangChain added a high-throughput, low-latency analytics layer that could support real-time insights, drill-downs, and efficient trace lookups while keeping deployment simple across local, self-managed, and cloud environments. By using ClickHouse sorting keys, ReplacingMergeTree, and materialized views, LangChain was able to scale LangSmith to production workloads and let users log large amounts of data, with the company explicitly crediting ClickHouse for making that possible.

LangChain

Ankush Gola

Co-Founder