Making it Fast - How Zotonic Maximizes Erlang's Power
summary
Zotonic explains how Erlang and mochiweb get applied end to end: compiled templates held in memory, a single template-compiler and single image-resize worker, two-level in‑memory caching with a site-wide dep cach and per-request cache, and combining identical computations across processes to avoid IO. Benchmarks show 7.5k req/s for a simple JSON endpoint and ~10 ms template render on a laptop, and a deployed site handles the election load—about 500k votes—with commodity VPSes.
AI-written summary · about
chapters (16)
- 0:00The scaling challenge
- 2:37Understanding web traffic
- 5:13Zotonic goals
- 7:55Core features
- 10:05Architecture and data flow
- 14:02Performance optimization
- 17:23Caching strategies
- 20:45Dependency-based caching
- 25:53Process dictionary caching
- 28:42Erlang and web machine
- 32:26Preventing bottlenecks
- 35:07Performance benchmarks
- 37:20High-traffic case study
- 38:44Nationwide voting project
- 42:53Zamo and future directions
- 45:27Conclusion
related talks
44mAre you being servered? - exploring a serverless web
32mCompilers: The Next Frontier in Web Performance
59mClosing Keynote
28mCompilers: The New Frontier In Web Performance
45mSpeeding up Code, Webpages and Teams
27mCache=Cash!
30mQwik Go For 100
25mMy Blog is Hilariously Overengineered to the Point People Think it's a Static Site