Serverless or Containers? Who Cares Pick Both!
summary
A pizza menu microservice is prototyped on functions (AWS Lambda, Azure Functions) then refactored to containers (ECS/Fargate, Azure Container Apps) after telemetry shows intermittent 1s API responses caused by cold starts; analysis finds cold starts <1% of invokes. Recommendation: run synchronous, latency-sensitive APIs as long‑lived containers and asynchronous workers as short‑lived functions; use ports and adapters to keep business logic portable; example routes 60–70% small files to Lambda and large files to ECS with the same container image.
AI-written summary · about
chapters (9)
- 0:00Software development complexity
- 3:14First microservice from monolith
- 12:19Serverless spectrum and compute
- 21:47Cold starts in serverless
- 30:14Ports and adapters design
- 40:00Sync vs async patterns
- 46:37Serverless or containers?
- 51:28Lambda and ECS example
- 53:36Q&A
related talks
41mContainers or Serverless Functions - When to choose what?
01h 00mMicroservices and Serverless
51mWho needs containers in a serverless world?
33mContainers and AWS Lambda: Choose the optimal compute for applications
35mContainers and AWS Lambda: Choosing optimal compute for modern apps
24mContainers vs Serverless from a DevOps standpoint
01h 02mServerless Architecture Patterns
48mServerless beyond functions