Case Study: GitLab achieves sub-second analytics at massive scale with ClickHouse

How GitLab serves sub-second analytics to 50 million users

GitLab needed a way to deliver sub-second analytics to its 50 million users across GitLab.com, GitLab Dedicated, and self-managed deployments. As its product analytics demands grew, Postgres could no longer handle the scale and performance requirements for user-facing analytics, prompting GitLab to evaluate a specialized OLAP database. Vendor name: ClickHouse, used as the core analytics engine.

GitLab implemented ClickHouse Cloud and OSS ClickHouse, along with its own ClickHouse operator and a hybrid data access layer to route workloads between Postgres and ClickHouse. With ClickHouse, queries that previously took 30–40 seconds on 100M+ rows now run in under a second, with one hierarchical query dropping to 0.24 seconds. ClickHouse also improved ingestion speed, reduced storage usage by nearly 10x in testing, and enabled features like Contribution Analytics, GitLab Duo, and SDLC trends at scale.

GitLab

Dennis Tang

GitLab