Software Operations and Human Cognition

on YouTube ↗43m

" summary

Argues that experts win by mental simulation and chunking not buzzwords, and shows concrete fixes: adding a script to capture thread dumps, reading them, finding an unconfigured database connection pool and increasing connections cures a months‑long brownout. Lists actual tools used—thread dumps, profilers, packet captures, tracing—and concrete frictions like slow rollout of debug builds and inaccessible SSH. Claims expertise reduces wasted work chasing red herrings, and that better introspection beats relying on noisy logs alone.

AI-written summary · about

" chapters (12)

1. 0:00Troubleshooting intro
2. 3:33Incidents vs ongoing issues
3. 8:12Real-world case study
4. 14:35Experience in troubleshooting
5. 17:03Logs and metrics limits
6. 20:57Introspection tools
7. 24:31Management’s role
8. 27:00Hidden dependencies
9. 30:55Cognitive science of troubleshooting
10. 34:29Recognition-primed decisions
11. 38:18Frame theory in systems
12. 41:07Key takeaways

" related talks

• 34mTaking Human Performance Seriously in SoftwareJohn Allspaw · DevOpsDays Seattle 2018
• 20mSeeing the Invisible: Discovering Operations ExpertiseJohn Allspaw · Velocity Conference Amsterdam 2015
• 18mSeeing the Invisible: Discovering Operations ExpertiseJohn Allspaw · Velocity Conference New York 2015
• 01h 05mThe art and methodology of troubleshootingAlexey Diomin · JavaDay Kiev 2014
• 25mHow to detangle and understand productionLiz Fong Jones · The Lead Developer New York 2019
• 31mClosing KeynoteAndrew Clay Shafer · DevOpsDays Istanbul 2019
• 38mLost Art of TroubleshootingLeon Fayer · DevOpsDays Boston 2017
• 01h 39mPostOps: A Non-Surgical Tale of Software, Fragility, and ReliabilityTodd Underwood · LISA '13