Case Study: Vimeo achieves petabyte-scale, high-performance search with Elastic

Vimeo - Customer Case Study

Vimeo, a high‑traffic video platform ingesting roughly 75 years of video per month and serving over 50 PB monthly, needed to deliver fast, reliable search at scale (over 1k search requests/second) across massive, frequently updated datasets. The challenge was designing indexes and a service layer that could handle rapid growth, expensive document updates, heavy bulk indexing, and major Elasticsearch version migrations without degrading user experience.

Vimeo implemented an Elasticsearch-based architecture with thoughtful index partitioning (time- and property-based indices like videos-public/videos-private), reusable query templates and DSL, and a dedicated search service (“Waldo”) with adapters and an A/B routing system (ABLincoln) to migrate versions safely. Combined with operational best practices for bulk indexing (disable refresh, set replicas to zero temporarily, throttle updates), this approach let them double the data stored in ES with no noticeable impact, sustain high query throughput, and perform iterative development and version upgrades smoothly.

Vimeo

Chris Simpson

Software Engineer