API Optimization Tale: Monitor, Fix and Deploy (on Friday)
summary
Migrating a billing client from REST to GraphQL, the team adds feature flags, fallbacks and custom request instrumentation that logs method, args, stacktrace, latency and errors. A bad parameter-slicing change causes full-table loads and a production outage with 20× CPU spike and 50× memory increases. Fixes include batching/preloading to eliminate N+1 calls, using GraphQL to fetch 4 instead of 40 fields, adding jitter and Sidekiq rate limiting, and choosing a quick DB-field lookup over a Kafka projection; deployments stay guarded by CI and easy rollback.
AI-written summary · about
chapters (6)
- 0:00Start
- 7:54EXTRACTION: GRAPHQL
- 12:35FLOOD OF REQUESTS
- 15:23FLOOD OF DB QUERIES
- 30:32FREQUENTLY NEEDED DATA
- 42:58NIHIL NOVI SUB SOLE
related talks
24mA GraphQL Survival Kit
28mPanel Discussion: The GraphQL Production Roundtable
31mEmbracing resiliency in GraphQL-Drop the cache-aside pattern and overcome the...
42mPanel - Fortune 100 Graphs - Powering Today's Need and Tomorrow's Trends
41m'Divide et impera' with GraphQL and SSR
49mBuilding and operating at scale with feature management
41mDemand Driven Applications with GraphQL by Vinci Rufus
26mThe Easy Way and the Hard Way: Blue-green GraphQL Deployments