Scalable and Resilient GraphQL Subscriptions with Distributed SQL Databases
summary
Demonstrates scaling GraphQL subscriptions to 1,000,000 subscribers end-to-end by using Hasura + YugabyteDB on Kubernetes, a custom GraphQL perf tool, and an e‑commerce order-tracking schema; live queries (polling) create auxiliary DB load the team measures. Cluster scales from 3 to 5 Yugabyte nodes and Hasura pods are increased without downtime; killing a tserver triggers follower→primary failover with only a brief two‑minute ops dip and transient latency spike, validating resiliency.
AI-written summary · about
chapters (22)
- 0:00Introduction
- 1:14GraphQL for UX Developers
- 2:17Scalability
- 3:18Resiliency
- 4:05Distributed SQL
- 5:15Connecting to any GraphQL Engine
- 6:07Demo
- 6:36Why GraphQL Subscriptions
- 7:12Results
- 8:48Demo Setup
- 9:23Sample Application
- 10:27Simple Subscription
- 10:46Setup
- 12:07Kubernetes Cluster
- 13:38GraphQL Perf Tool
- 14:52Auxiliary Traffic
- 16:24New Orders
- 17:06New Orders Simulation
- 18:49Yucabit
- 20:33Cloud Native Database
- 21:07Data Node Failure
- 22:36Outro
related talks
19mBuilding Planet Scale Cloud Native Apps with GraphQL & Distributed SQL
08mScalable GraphQL Applications Powered by Cloud-ready Distributed SQL Database
27mAn API First Approach to Adopting a Distributed SQL Database
28mModernizing App Development: Planet-scale GraphQL & Distributed SQL
31mA Migration Journey from Amazon DynamoDB to Yugabyte YSQL and Hasura
29mEnterprise Scale: How a Distributed Architecture Can Deliver Scale and Resilience
30mBest Practices for GraphQL and GraphQL Subscriptions at Scale
25mHow Sharekhan Scaled Trading Platform with Yugabyte's Distributed SQL