Thayne McCombs Interview
summary
Company moves to continuous delivery and faces test coverage, backward-compatibility and release-safety gaps while shipping more often. Relational data is being sharded by application boundary with separate databases per service, which reduces vertical scaling but adds operational complexity and coupling that causes cascading failures. All services live in a single AWS region now; plans include multi-region deployment and geographic data sharding. Backend uses many threads and adopts functional programming with immutable data pipelines to reduce data races.
AI-written summary · about
related talks
35mBuilding Systems that Last
24m"Misadventures in Continuous Delivery"
31mDevOps in the Cloud: Case Studies of Amazon.com teams and their resilient architectures
27mMicro-transformation
40mUncoupling Systems
01h 03mMicroservices: Decomposing Applications for Testability and Deployability
44mA Software Engineer's Journey Through Microservices Data Sharing
48mMoving from spaghetti infrastructure to composable cloud environments
