You Don't Want Serverless

on YouTube ↗49m

" summary

Argues that serverless (AWS Lambda, Azure Functions, Cloudflare Workers) leaks complexity: database connection pooling breaks by default causing exhausted connections unless using PG Bouncer or RDS Proxy; cold starts force AOT or provisioned fixes like provisioned concurrency; observability and testing require code changes; account-level concurrency limits throttle throughput; metered chains can create runaway bills (example of a deployment charging $100k bill); Amazon Prime Video cuts cost 90% by moving off functions to VMs.

AI-written summary · about

" chapters (17)

1. 0:00Introduction
2. 2:24Defining serverless
3. 4:54Main issues overview
4. 6:29Simplicity and state problems
5. 11:16Runtime and security challenges
6. 13:35Developer experience pitfalls
7. 15:36Testing strategies
8. 18:05Platform churn and environment limits
9. 19:43Observability and architecture
10. 22:17Orchestration and cost loops
11. 25:44Cold starts and warming
12. 27:56Code optimization techniques
13. 30:23Vendor incentives and lock-in
14. 33:27Concurrency limits and costs
15. 38:00Financial risks and total cost
16. 42:10When to use serverless
17. 47:30Conclusion

" related talks

• 51mConfusion In The Land Of The ServerlessSam Newman · JavaDay Ukraine 2017
• 52mConfusion In The Land Of The ServerlessSam Newman · NDC Sydney 2017
• 55mConfusion In The Land Of The ServerlessSam Newman · NDC Oslo 2017
• 01h 00mMicroservices and ServerlessSam Newman · NDC DevOps 2020
• 28mServerless - the way microservices should beIvan Jovanovic · OdessaJS 2018
• 01h 00mServerless - reality or BS - notes from the trenchesLynn Langit · NDC Oslo 2017
• 15mThe Ops in ServerlessJennifer Davis · SREcon19 Americas
• 31mServerless Rails: Understanding the pros and consDaniel Azuma · RailsConf 2021