Confusion In The Land Of The Serverless
summary
Argues serverless is essentially a higher‑level platform as a service with five practical criteria (no host management, autoscale, precise billing, non‑host performance, implicit HA) and shows real failures: Lambda-style scaling breaks per‑client throttles so connection pools vanish and a company’s Lambdas overwhelm Redis—hitting 10,000 connections per node—unless rearchitected. Notes container isolation risks, migration as an opportunity cost (avoid DynamoDB if portability matters; prefer RDS), and recommends treating functions as event-driven building blocks while planning hybrid resiliency and throttling.
AI-written summary · about
chapters (11)
- 0:00Introduction and confusion
- 4:40Defining serverless
- 8:34Characteristics of serverless
- 13:05Context within cloud computing
- 17:17The concept of heavy lifting
- 20:32Resiliency challenges
- 27:56Throttling and state management
- 35:32Security implications
- 38:35Lock-in and migration costs
- 43:24Conclusion and reflections
- 48:50Q&A session
