Serverless Python in the real world: common patterns & challenges

on YouTube ↗50m

" summary

Built a serverless URL shortener that records clicks and redirects, deployed on AWS Lambda + API Gateway using frameworks like Serverless and Zappa (Flask adaptor). Load tests up to ~1,000 concurrent invocations show latency spikes after ~250 concurrent cold starts; packaging native binaries required Docker builds to avoid runtime failures. Hitting Postgres connection limits forces either a proxy like pgbouncer or a switch to Aurora Serverless / queueing (SQS). Pricing note: first million Lambda invocations free, then roughly $0.20 per million — works for bursty loads, not sustained high throughput.

AI-written summary · about

" chapters (12)

1. 0:00Introduction
2. 7:04Product overview
3. 10:18AWS Lambda basics
4. 13:33Scalability and dependencies
5. 17:09Zappa framework solution
6. 21:37Binary system challenges
7. 24:36Testing and error handling
8. 28:52Monitoring and logging
9. 31:50Local testing and mocking
10. 34:24When to use serverless
11. 36:12Serverless ideology
12. 38:30Q&A

" related talks

• 31mA quick intro to AWS ServerlessCurtis Maloney · PyCon AU 2016
• 34mPython in Serverless ArchitecturesDaniel Lindeman · PyOhio 2018
• 19mBuilding Serverless API BackendsRich Jones · Nordic APIs Platform Summit 2016
• 54mServerless Python on AWS Lambda using the Serverless FrameworkTed Mathew dela Cruz · PyCon APAC 2019
• 01h 00mServerless - reality or BS - notes from the trenchesLynn Langit · NDC Oslo 2017
• 42mNo Server Is Better Than No Server: Building Blocks for Serverless ArchitecturesConstantin Gonzalez · code.talks 2017
• 29m2 Years, 0 Servers: Lessons learned from running a startup 100% on ServerlessTal Bereznitskey · Cloud Native & OSS Day Tel Aviv 2019
• 25mA Crash Course on Serverless Applications in PythonJames Saryerwinnie · PyGotham 2018