This document discusses serverless architecture and the Python ecosystem for serverless applications. It introduces serverless computing and its benefits like automatic scaling and pay per use billing. Popular serverless platforms like AWS Lambda are examined along with their limitations. Frameworks for Python serverless development are presented, including the Serverless Framework for deploying AWS Lambda functions and Zappa for deploying WSGI applications on Lambda. The talk concludes that serverless architecture is the next evolution of cloud computing and an opportunity for Python developers.

1. Python In The Serverless Era
Benny Bauer
Software Architect, Autodesk
benny.bauer@autodesk.com
@benikbauer

2. ▪ Hello
▪ Cloud Evolution
▪ Serverless Architecture
▪ Ecosystem
Agenda

3. Hello

4. Hello

5. AutoCAD 360

6. AutoCAD 360 Backend

7. Cloud Evolution

8. It’s Evolution Baby!
Icons made by freepik from www.flaticon.com

9. ▪ Hosting
▪ Data access
▪ Authentication
▪ Notifications
▪ Monitoring
▪ Analytics
Backend as a Service

10. Serverless
Architecture

11. Introducing Serverless

12. ▪ Fully managed compute
▪ Provisioning, patching
▪ Scalability
▪ Monitoring
▪ Logging
▪ No ops
▪ Just deploy your code
▪ Pay only for actual usage ==
Full utilisation!
Serverless Architecture

13. AWS Lambda - How it works
origin: https://aws.amazon.com/lambda/

14. ▪ REST API
▪ Stateless services
▪ Suitable for Slack apps (though not bots)
▪ Events
▪ File processing (S3 event) & Data ingestion (Kinesis event)
▪ Incidents handling (CloudWatch event)
▪ IoT
▪ Scheduled tasks
▪ Monitoring, sanity tests, load testing
▪ Periodical jobs
Use cases

15. ▪ Stateless
▪ Autoscaled according to demand (events or requests)
AWS Lambda Characteristics

16. ▪ Need to keep it warm
▪ Convenience can lead to vendor lock-in
▪ Languages: Python 2.7, Node.js, Java 8
▪ Execution time is limited to 5 min
▪ Concurrent execution is limited to 100*
▪ Various payload and disk size limits
▪ Not in all AWS regions
▪ No SSH (which is good!)
AWS Lambda Limitations

17. Code should be:
▪ Small
▪ Short-lived
▪ Stateless
The Required Mindset

18. Ecosystem

19. AWS Lambda & API Gateway are nice, but...
▪ Configuration headache
▪ IAM roles
▪ API Gateway
▪ Deployment headache
▪ Packaging
▪ Uploading
▪ Rollback

20. Serverless Framework
Application framework for building applications
exclusively on AWS Lambda and API Gateway
Serverless
origin: https://github.com/serverless/serverless

21. Serverless Framework
▪ Wraps AWS Lambda, API Gateway and IAM
▪ Manages deployments
▪ CLI
▪ Modular, functions can be easily shared
▪ Extensible via plugins
▪ Thriving community
▪ Node.js - posterboy, Python - foster child

22. $ sls project create

23. $ sls function create

24. ▪ handler.py
▪ Entry point to your implementation
▪ s-function.json
▪ Endpoints configuration
▪ Events configuration
▪ Env vars definition
▪ Lambda configuration (runtime, timeout, size, etc.)
# implement
functions
|__function1
|__event.json
|__handler.py
|__s-function.json

25. $ sls dash deploy

26. Zappa
Python WSGI applications deployment on AWS
Lambda + API Gateway.
Rich Jones
Miserlou
origin: https://github.com/Miserlou/Zappa

27. Zappa Architecture

28. Zappa example

29. Zappa example

30. Zappa - How it works
$ zappa deploy <env>
1. Zips code and dependencies
2. Create AWS Lambda and deploys the zip
3. Creates endpoint on API Gateway and ties to AWS
Lambda

31. Serverless architecture is the next generation
of cloud evolution
Takeaways

32. The Serverless ecosystem is on the rise,
many interesting opportunities for
the Python community to contribute!
Takeaways

33. ▪ awesome-serverless
▪ https://github.com/Miserlou/Zappa
▪ Serverless Framework talk (aka JAWS) on AWS re:invent
https://youtu.be/D_U6luQ6I90
References

34. Questions?

35. Thank You!