Rebuilding Twitter’s public API
summary
Twitter builds a GraphQL-based multi-tenant platform (Scala, OpenAPI Spec, thrift) to host v2 endpoints, consolidating fragment-driven field configs and persistent queries so endpoints are generated from reusable resource fields and selections. Business logic lives in StrataQL-managed columns or thrift services, optional fields (e.g. public_metrics) are included via GraphQL includes, and SLOs with PagerDuty/Grafana monitoring enforce ownership. This replaces scattered HTTP services and enables faster, consistent API rollout across many endpoints.
AI-written summary · about
chapters (7)
- 0:00Start
- 12:40Selection Config
- 12:53Selection: Single Tweet
- 14:39Selection: Search Tweet
- 17:16Generated Platform Tweet Fragment
- 18:19Tweet By Id W/ Public Metrics
- 18:55Tweet Search W/ Public Metrics
related talks
24mStrato: Twitter’s Virtual Database Powered by Microservices
25mGraph-First Services Using GraphQL
28mThe Evolution of Twitter's Edge
22mThe Open Source Technology Behind a Tweet
22mDelivering Exceptional Infrastructure as Code through REST and GraphQL
23mAugmenting a Legacy REST API with GraphQL
27mFixing Twitter: Improving the Per
42mScaling to Billions of Real-time Recommendations Daily
