How shit works: TCP/IP
summary
Explains how TCP turns an unreliable packet network into a reliable, ordered byte stream using a three-way handshake for connections, RTT sampling to set retransmission timers, a sliding window for flow control, MSS/MTU negotiation, retransmission on timeout, and fast retransmit via triple duplicate ACKs; shows stop‑and‑wait is unusably slow, uses a 4‑byte example, notes RTO defaults (historically 1–3s) and TIME_WAIT (≈2×MSL, ~1 minute), and warns against SO_LINGER hacks.
AI-written summary · about
chapters (20)
- 0:00Intro
- 1:44In Theory...
- 3:08In Practice
- 5:22What Worked?
- 7:53So... What Went Wrong?
- 9:45Slight Digression
- 11:58Full Disclosure
- 13:37Product Management 101
- 15:30Proposed Solution
- 17:24Product Management 102
- 19:07Point-to-Point
- 21:22Connection-Oriented
- 26:31TCP Flow Control
- 27:42Sliding Windows
- 31:21Packet Loss
- 36:53TCP Fast Retransmission
- 39:25TCP Dynamics
- 46:18Closing the Connection
- 49:00Now It Gets Interesting
- 51:55Further Reading
related talks
53mHow shit works: TCP/IP
23mThe Short and Happy Lives of TCP and HTTP Requests
23mACK to the Future: TCP in 2025
43mHow internet congestion control actually works in the bufferbloat age
45mReducing Network Latency: Innovations for a Faster Internet
15mHA/TCP: A Reliable and Scalable Framework for TCP Network Functions
16mR-TCP: A Framework to Optimize TCP Performance Over Rate-Limiting Networks
34mDeterministic Networking for Real-Time Systems (Using TSN)