Case Study: Reverb Technologies achieves 10x lower API costs and millisecond-scale performance with MongoDB

Reverb Technologies - Customer Case Study

Reverb Technologies runs the world’s largest English-language “live” dictionary, mapping words from real-world text sources and a public API to reveal usage in context. The company needed to ingest and query massive, continuously refreshed text streams (3.5 TB, 20 billion records) for real-time analysis but hit severe performance and availability limits on MySQL—table locks, outages, and scaling bottlenecks.

Reverb migrated its entire corpus to MongoDB in a month with one developer and zero downtime, adopting replica sets for high availability. MongoDB now powers 20 million API calls per day, sustains huge insert and query throughput (average 500k requests/hour, 8k inserts/sec with bursts to 50k), eliminated the memcached layer, and simplified the stack. The change cut cost per API request 10x, reduced code by 75%, and dramatically improved response times (example fetches from 400ms to 60ms; entries from 20ms to 1ms).

Reverb Technologies

Tony Tam

Vice President of Engineering and Technical Co-founder, Reverb Technologies