Web Compression: How to (Re)Take Control?
summary
Covers web response compression: explains algorithms (Huffman, LZ77, deflate), gzip headers/limits, and streaming decompression during HTML parse. Shows measurable performance wins (e.g., a Wikipedia-like page load on 3G drops from ~17s to ~7s) and compares codecs (brotli typically saves 8–12% vs gzip), recommending precompressing static assets with brotli level 11 or zopfli at build time and using moderate levels (~5–6) for on-the-fly compression; also notes Chrome support for zstd and shared dictionaries.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 1:06Minification vs compression
- 3:35Real compression gains
- 5:02Performance impact
- 6:26Wikipedia demo
- 7:48Alarming stats
- 8:51HTTP compression protocol
- 10:58Streaming load flow
- 13:49Gzip history
- 15:15Huffman coding
- 20:34LZ77 algorithm
- 25:37Algorithm evolution
- 28:47How gzip works
- 34:49Compression levels
- 36:36Brotli benefits
- 40:10Pre-compressed files
- 41:58Shared dictionaries
- 42:40Zstandard in Chrome
- 44:47Best practices summary
related talks
51mWeb Compression: How to (Re)Take Control?
46mNot Only Gzip
42mHoney, I Shrunk the Bits. Or how modern HTTP compression works.
52mCompression Dictionaries
03mWhen small is better: Brotli compression in a real case
15mBeyond Gzipping
36mBreaking The Bloat: Achieving The Fastest Navigations on The Web
24mknock-knock-who-there-file-compression-TALK-FINAL-4.tar.trz.bz2.gz