Scaling GraphQL on AWS: Production Architecture for High-Volume Data Systems
summary
Production GraphQL runs at >=1 billion ops/day across hundreds of clients using a self-hosted layer with AWS AppSync-compatible patterns, Lambda, DynamoDB and SNS/SQS for eventing; clients supply a JSON config-driven federation that controls routing, per-section caching, alarms and storage migration. The stack enforces DLQs and recommends FIFO queues to avoid duplicates, uses composite partition keys to mitigate DynamoDB hot partitions, separates client vs platform metrics, and cuts onboarding from weeks to hours with LLM-assisted onboarding and automated backward-compatibility tests.
AI-written summary · about
chapters (13)
- 0:00Introduction to GraphQL scaling
- 2:34GraphQL limitations in production
- 4:20Config-driven federated architecture
- 6:05Access and injection strategies
- 7:30Metrics isolation and client control
- 9:59Event-driven data delivery
- 12:06Layered caching strategies
- 13:52Scaling with partitioned storage
- 14:33Designing for failure and errors
- 16:20Backward compatibility challenges
- 17:24Key lessons and takeaways
- 18:06Recent advances in onboarding
- 19:31Q&A
related talks
28mGraphQL at scale with AWS
34mGoing Serverless with GraphQL
31mLessons from Scaling GraphQL to Half a Billion Requests per Minute
28mA Real-World GraphQL Application in Production
38mAPI Modernization with GraphQL
36mDeveloping Serverless Applications with GraphQL
34mPower real-time serverless applications with GraphQL and AWS AppSync
17mBuild cost effective Asynchronous GraphQL APIs using Serverless.