Python: Ludicrous mode (with Django)
summary
Latency is primarily a function of distance, not CPU or bandwidth: examples show 36 GB via carrier pigeon — 20 minute RTT -> 30 minutes total and NBN 1 Mbps -> 10 hours for 36 GB. Recommendations target real fixes: use Django streaming responses to send rows as produced, prefer locked in-process caches and template fragment caching, apply select_related/prefetch_related, annotate/aggregate, watch queries with Django Debug Toolbar or New Relic, avoid cross-continent cache servers, and New Atlas uses Postgres xmin for cache invalidation and proposes streaming template responses and HTTP/3 late status codes.
AI-written summary · about
chapters (12)
- 0:00Introduction and context
- 1:32Latency versus bandwidth
- 2:54Latency scenarios
- 5:36Web protocol overhead
- 7:44Django request processing
- 10:22Streaming responses
- 13:05Caching strategies
- 15:17Template optimizations
- 16:56Lazy evaluation
- 19:10Database query optimization
- 20:45Recap and optimization workflow
- 23:27Conclusion and Q&A
related talks
28mThe life of a web request - techniques for measuring and improving Django application performance
37mPerformant Django
42mA Nice Problem to Have: Django Under Heavy Load
45mCutting latency in half: What actually worked—and what didn’t
17mOff the Beaten Path: Exploring Unusual Roads to Django Performance
47mReal World Django Q&A
45mCaching: A trip down the rabbit hole
47mReal World Django Q&A