The troubles with distributed systems chapter

Designing Data-intensive Applications with Martin Kleppmann

Topic 8 of 14

Published Apr 22, 2026 · Clip 42:48–46:53 (4:05)

Martin describes the chapter's purpose: defending the theoretical models of distributed systems (unbounded network delays, crashes, unreliable clocks) with real-world case studies drawn from published postmortems. The moral is that failures are not rare and reliability requires handling weird edge cases; he shares anecdotes like sharks biting undersea cables and cows stepping on land cables. He notes the chapter informs rather than prescribes: risk-versus-cost decisions belong to the business.