Case Study: Algolia achieves sub-800ms search analytics and 5–10B rows/day ingestion with Citus Data

With Citus, Algolia Speeds Up Search Analytics & Eradicates Database Management Headaches

Algolia is a hosted search-as-a-service used by thousands of websites and apps that also provides analytics on user search behavior. Their self-hosted Elasticsearch analytics stack struggled with frequent slowdowns, crashes, and costly maintenance as traffic spiked (including billions of search operations per day), so Algolia needed a scalable, easy-to-manage relational solution that could deliver real-time query performance without ballooning costs.

Algolia migrated to Citus (a managed distributed Postgres), using sharded data, five‑minute and daily rollup tables, and Postgres extensions like TopN and HyperLogLog to precompute metrics and enable fast aggregations. The Citus solution ingests 5–10 billion rows per day (benchmarked up to 7M rows/sec), drives 95% of queries under 800 ms, eliminated major downtime and heavy on-call load, and freed Algolia to focus on product development and new analytics features.

Algolia

Sylvain Friquet

Software Engineer