Case Study: Spotify achieves real-time, always-on playlists at 40,000 requests/second with DataStax

Spotify supports 40K requests/second to deliver the right music to their customers fingertips

Spotify, a global music streaming service with over 40 million monthly users and a catalog of 24+ million songs, needed to deliver personalized playlists and real-time recommendations at massive scale. As usage exploded, PostgreSQL could not meet Spotify’s performance, multi–data-center availability and high-write requirements, so the company sought a highly available, horizontally scalable datastore.

Spotify adopted Apache Cassandra’s masterless architecture to power product catalogs, playlists and other customer-facing services, integrating it with Hadoop/Spark for analytics. Today 500+ Cassandra nodes across four data centers support 40,000+ requests/second, 1.5 billion+ playlists and tens of terabytes of compressed data, delivering low-latency, highly available service and strong confidence in data durability.

Spotify

Axel Liljencrantz

Backend Engineer