The Short and Happy Lives of TCP and HTTP Requests
summary
Follows a single HTTP request from browser to OS to router to server and back, showing why latency dominates over bandwidth (speed‑of‑light sets a lower bound — ~37 ms theoretical min NYC–London, real ~70–90 ms), why new TCP connections cost an RTT (handshake + slow‑start can add ~10 round trips — e.g. 250 KB costing ~400 ms at 40 ms RTT), why cookies bloat every request, why modern pages make ~112 requests per page, and why workarounds are used: keep‑alive tuning, CDNs, concatenation, domain sharding and SPDY/HTTP2.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:59Understanding the OSI model
- 3:24The impact of latency
- 7:03Bandwidth and page loads
- 8:59TCP overhead and slow start
- 12:58Optimizing connections
- 15:32Headers and asset reduction
- 18:38Advanced loading techniques
- 21:16Future trends and resources
related talks
49mHTTP: History & Performance
52mDeath, Taxes, and HTTP
28mHTTP: History and Performance
16m"Latency: Why You Should Worry and What You Can Do About It"
36mMessing with JavaScript and the DOM
15mYou and Me Learn All About HTTP
43mPerformance in the World of HTTP 2.0
40mWhirlwind tour through the HTTP2 spec