DjangoCon 2008 Keynote: Cal Henderson
summary
Built the original Flickr and runs a site serving 2.8 billion photos at peaks of 35,000 photos/second, argues Django is great for getting started but missing production features needed at web scale. Lists concrete gaps: no easy master‑slave replication and failover handling, no automatic sharding or global unique ID strategy, weak denormalization tooling, heavyweight sessions, no parallel request helpers, crude SQL translation that breaks indexes, clumsy migrations and deployments, and poor query/debug tooling. Offers workarounds used at Flickr (PHP, MySQL, memcached, manage.py scripts).
AI-written summary · about
chapters (15)
- 0:00Introduction
- 5:16Built blogging engine in Django
- 7:06Python and frameworks problems
- 11:03Python language flaws
- 16:21Framework limitations
- 21:56Scaling Django challenges
- 29:19Database sharding and ID conflicts
- 35:02Denormalization and performance
- 37:50High performance and sessions
- 43:27Maintainability and templates
- 48:46Debugging tools and SQL
- 51:35Django 1.0 release
- 52:40Pluralization and models
- 56:11Deployment and migrations
- 57:36Final thoughts and Q&A
related talks
47mHigh Performance Django at Ten: Old Tricks & New Picks
47mBuilding Scalable Web Apps
29mHow Django helped us make a better conference website
52mChallenges in developing a large Django site
38mdjango-cms: Friends don't let friends use Drupal.
20mE-Commerce with Django at Scale...
22mDjango Channels Teaching a mature framework new tricks
29mSwiss Army Django: Small Footprint ETL