Case Study: Twitch achieves 2M+ concurrent viewers with scale, simplicity and high availability with Redis Labs' Redis Enterprise Cluster

Enabling 2M + Concurrent Users at Twitch with Scale, Simplicity and High Availability

Twitch, the world’s leading social video platform, serves 100M+ monthly users and more than 2M concurrent viewers, with chat rooms that can host hundreds of thousands of participants. Their engineering challenge was to provide an extremely high-throughput, low-latency persistent data store for site-wide chat while keeping operations simple, highly available and reliable.

Twitch adopted Redis Labs Enterprise Cluster as a managed Redis solution (with VPC peering), eliminating client-side sharding and much of the operational overhead and allowing nearly instant provisioning of Redis instances. The move—also extending Redis use from caching to token caching and view counting—delivered stable high performance, true high availability and minimal engineering effort, boosting confidence in Twitch’s ability to handle extreme scale.

Twitch

Daniel Lin

Twitch Staff Engineer