10 case studies
What actually broke. And how they fixed it.
Post-incident narratives from real engineering teams. Each case names the architectural assumption that failed, the user-visible symptom, and the decision the team made under pressure.
- cloudflare-workers-cold-startCase
Case: Cloudflare Workers Chose V8 Isolates Over Containers
Read case → - discord-message-storageCase
Case: Discord's Move from Cassandra to ScyllaDB for Trillions of Messages
Read case → - figma-multiplayerCase
Case: Figma's Multiplayer Server, CRDT-Inspired but Not Quite
Read case → - instagram-feed-fanoutCase
Case: Instagram's Early Architecture, PostgreSQL Plus Pragmatic Caching
Read case → - netflix-chaos-engineeringCase
Case: Netflix Built Chaos Monkey to Force Resilience by Default
Read case → - shopify-storefront-renderingCase
Case: Shopify's Storefront Stack, Liquid Plus Hydrogen
Read case → - slack-message-orderingCase
Case: Slack's Real-Time Messaging and Channel Servers
Read case → - stripe-idempotency-keysCase
Case: Stripe's Idempotency-Key Design
Read case → - uber-h3-geo-indexingCase
Case: Uber Built H3 to Index the World as Hexagons
Read case → - youtube-vitessCase
Case: YouTube Built Vitess to Shard MySQL Without Throwing MySQL Away
Read case →