Deploy Python apps in 5 min with a PaaS
6 likes6,556 views
The document outlines the concept of Platform-as-a-Service (PaaS) for deploying Python applications efficiently, contrasting it with traditional infrastructure methods. It discusses the advantages of using PaaS, such as reduced operational overhead and ease of scaling, while listing various providers like Heroku, OpenShift, and Dotcloud. Additionally, it highlights configurations for Django applications on different PaaS platforms and considerations for when not to use PaaS.
![Configuring a server
• deciding what size to get (memory, disk)
• ge]ng all the dependencies installed on it
• SSHing into the machine to deploy stuff
(feels dirty, but he’s under Tme pressure)
1 day 1 day](https://image.slidesharecdn.com/deploypythonappsin5minwithapaas-130226132656-phpapp02/85/Deploy-Python-apps-in-5-min-with-a-PaaS-9-320.jpg)
Deploy Python apps in 5 min with a PaaS
- 1. Deploy Python apps in 5 minutes with a PaaS Nate Aune PyCon on the Charles, part 1 Feb. 25, 2013
- 3. 2010-‐2011 • Nothing equivalent to Heroku for Django developers • Built my own PaaS (“How hard can it be?”) • Shut down last year. Increased compeTTon from big corps meant a race to the boUom
- 5. Agenda • Why should I care? • What is a PaaS? What are the advantages? • Which PaaS should I use? • When might I not want to use a PaaS?
- 6. Where are you deploying/hosTng today? • Shared hosTng (i.e. WebfacTon) • Running your own servers (co-‐located)? • Using an IaaS provider (i.e. AWS or Rackspace) • Already using a plaZorm-‐as-‐a-‐service (PaaS)
- 7. Meet Dave
- 9. Configuring a server • deciding what size to get (memory, disk) • ge]ng all the dependencies installed on it • SSHing into the machine to deploy stuff (feels dirty, but he’s under Tme pressure) 1 day 1 day
- 10. Maintaining the server 1 day 1 day 1 day
- 11. Security 1 day 1 day 1 day 1 day
- 12. Scaling 1 day 1 day 1 day 1 day 1 day
- 13. Oh sh*t moment.
- 14. Sysadmin $80k te sed DBA $90k Train new guy 6
- 15. Why touch servers if you don’t have to?
- 16. PaaS Is it the promised land?
- 17. What is a PaaS? Platform-as-a-service enables developers to create innovative applications without operational overhead around configuration, deployment and management.
- 18. Layers of infrastructure Source: EngineYard “PaaS - State of the Market Survey” May 2012 - http://venturebeat.com/2012/07/05/engine-yard-paas-infographic/
- 19. Source: EngineYard “PaaS - State of the Market Survey” May 2012 - http://venturebeat.com/2012/07/05/engine-yard-paas-infographic/
- 20. Source: EngineYard “PaaS - State of the Market Survey” May 2012 - http://venturebeat.com/2012/07/05/engine-yard-paas-infographic/
- 21. Source: EngineYard “PaaS - State of the Market Survey” May 2012 - http://venturebeat.com/2012/07/05/engine-yard-paas-infographic/
- 22. Source: AppFog “Evolution of the Cloud: Toward a NoOps World” Jan 2012 http://gigaom.com/cloud/why-2013-is-the-year-of-noops-for-programmers-infographic/
- 23. Efficient, Elastic, Secure • Lots of applications co-located on a few servers • Drastically reduces resources • Add/remove servers depending on load • All secured using SELinux or LXC
- 24. Let them do the boring stuff • Patches and updates • Migrating applications • Backups / snapshots • Configuring everything (web servers, load balancers, modules, databases)
- 25. Define OS dependencies www: type: python systempackages: - openoffice.org - mysql-client-5.1 5 min
- 26. Backups 5m 5m https://devcenter.heroku.com/articles/heroku-postgres-fork
- 27. SSL 5m 5m 5m
- 28. Scaling 5m 5m 5m 5m https://openshift.redhat.com/community/developers/scaling
- 29. Caching 5m 5m 5m 5m 5m
- 30. IaaS vs. PaaS IaaS -‐ days PaaS -‐ minutes
- 31. Which PaaS? • Hosted • DIY • Hybrid (public with opTon to go private)
- 32. Who are the players? • CloudFoundry • Heroku (open source by VMWare) (now Salesforce) • OpenShift (Redhat’s open source PaaS) • AppFog • Dotcloud • App Engine (Google) • Stackato (ActiveState) • Gondor (Python only) • Elastic Beanstalk (Amazon) • Azure (Microsoft)
- 33. Who are the players? We’ll look at these ones tonight. • CloudFoundry • Heroku • OpenShift (open source by VMWare) (now Salesforce) (Redhat’s open source PaaS) • AppFog • Dotcloud • App Engine (Google) • Stackato • Gondor (ActiveState) • Elastic Beanstalk (Python only) (Amazon) • Azure (Microsoft)
- 34. Dotcloud polyglot from the start, very flexible, most Python centric
- 37. Anatomy of a Django app on Dotcloud customized settings.py for Dotcloud createdb.py to create the database dotcloud.yml to store config info mkadmin.py to make the admin user nginx.conf to config URL rewriting postinstall to run syncdb, collectstatic wsgi.py to serve using uWSGI
- 38. Heroku
- 41. Anatomy of a Django app on Heroku customized settings.py for Heroku Procfile to configure process requirements.txt to define dependencies
- 42. Stackato by ActiveState Python 3, Run anywhere, New Relic integration http://appsembler.com/blog/django-deployment-using-stackato/
- 45. Anatomy of Django app on Stackato customized settings file stackato.yml to define services wsgi.py to serve using uWSGI
- 46. OpenShift by Redhat Open source, Auto-scaling, Jenkins builds http://appsembler.com/blog/django-deployment-using-openshift/
- 48. Action hooks for running commands during build, deploy, post-deploy, etc. /data/ dir to store uploaded media files Anatomy of an OpenShift repo .htaccess to serve up static files application inside of wsgi dir setup.py instead of requirements.txt
- 49. Stackato OpenShift Dotcloud Heroku Python 2.7, 3.2 2.6 (2.7) 2.6.5, 2.7.2, 2.7.2 stackato runtimes 3.1.2, 3.2.2 PostgreSQL 9.1 8.4 9.0 9.1.6 MySQL 5.5 5.1 5.1 (Yes, via RDS) Persisted FS Yes Yes Yes (Yes, via S3) Redis Yes, 2.4 No Yes, 2.4.11 (Yes, via addon) MongoDB Yes, 2.0 Yes, 2.2 Yes, 2.2.1 (Yes, via addon) Memcached Yes, 1.4 No Yes (Yes, via addon) RabbitMQ Yes, 2.4 No Yes, 2.8.5 (Yes, via addon) Solr No No Yes, 3.4.0 (Yes, via Websolr) Cron Yes Yes Yes Yes Extensible Yes, apt-get install Yes, DIY cartridge Yes, custom service Yes, buildpacks WebSockets Yes Yes Yes Yes, via Pusher add-on
- 50. Stackato OpenShift Dotcloud Heroku Memory Configurable 1.5GB Unlimited RAM sandbox Apps Configurable No Yes, 2.8.5 (Yes, via addon) Solr No No Yes, 3.4.0 (Yes, via Websolr) Cron Yes Yes Yes Yes Extensible Yes, apt-get install Yes, DIY cartridge Yes, custom service Yes, buildpacks WebSockets Yes ? Yes Yes, via Pusher add-on
- 51. Why not PaaS? • Already invested in your own infrastructure. • Need to run on servers outside U.S. • Special requirements not met by PaaS services
- 52. OpenShid Pricing
- 53. Dotcloud pricing
- 54. Heroku pricing
- 55. Other resources • Wrap-up from PaaS bake-off http://appsembler.com/blog/wrap-up-from-paas-bake-off/ • Django deployment using PaaS http://appsembler.com/blog/django-deployment-using-paas/ • django-deployer https://github.com/natea/django-deployer • paasbakeoff - code examples https://github.com/appsembler/paasbakeoff/