Lessons from Scaling GraphQL to Half a Billion Requests per Minute

Session description

Learn about the scaling challenges we faced in terms of infrastructure, organizational growth, and the technological aspects of GraphQL usage within a hyper-growth startup. In September 2016, when our team of ten-odd full-stack engineers adopted GraphQL, we were handling a few thousand requests per second on our GraphQL layer. Fast forward to 2023, the engineering team has grown to 1,000 members managing half a billion requests per minute at peak. In this talk we aim to address critical questions regarding GraphQL's performance, when is the right time to integrate it, exploring whether GraphQL is predominantly a front-end abstraction or if it can also be effectively utilized in back-end operations for service-to-service communication. Drawing from years of extensive GraphQL usage at a significant scale—without the need for federation—we will introduce for the first time an open-source project that helped us elegantly design, streamline, and scale optimal GraphQL APIs.