Case Study: StoryFire achieves low-latency, scalable real-time queries for its social video platform with Rockset

StoryFire - Scaling a Social Video Platform on MongoDB and Rockset

StoryFire is a social video platform that helps creators share and monetize stories and videos; to power its social feeds and leaderboards it uses Rockset to index data from its transactional MongoDB Atlas system. Rapid user growth and influencer migrations created spikes in activity, producing a hot spot of read‑intensive aggregation and join queries that required low-latency responses and high concurrency without starving MongoDB transaction workloads.

Rockset was deployed as a managed “speed layer” connected to MongoDB Atlas via CDC, providing automatic converged indexes, Query Lambdas for RESTful SQL access, and zero‑downtime scaling of compute. By offloading expensive joins and aggregations to Rockset, StoryFire saw roughly flat query response times when increasing lookback windows from 30 to 90 days, predictable millisecond‑range performance (double‑digit ms for typical users), and proportional performance gains as compute was increased in tests at 4, 8 and 16 vCPUs—reducing load on MongoDB and preserving transactional performance.

StoryFire

Brian Spitz

CEO and Co-founder