With over 150 million users viewing and booking over 6 million listings worldwide, Airbnb’s technical systems need to operate at high speed and massive scale. In this presentation, we’ll share how we built Airbnb’s Read-Optimized Stores (ROS), a data framework designed to improve performance and reliability for read-heavy workloads. The ROS framework powers over 50 materialized views across Airbnb, processing billions of events and writing hundreds of millions of documents daily, providing a robust solution for data access and enabling faster product iteration. You’ll hear about the technologies we used, specific choices we made, and what you should consider when building a datastore for performance at scale.