DevOps Patterns and Antipatterns for Continuous Software Updates
summary
Explains why updates must be tiny, frequent and safe with concrete failures: Java moves to six‑month releases but 75% of users stay on Java 8; a router OTA update reset configs and cut home internet; Knight Capital forgets to update one of eight servers and loses about $400M; Cloudflare misconfigured regexes twice (2019 and 2020) and knocks large regions offline. Recommends automation, progressive delivery, observability (tracing/logging/monitoring), feature flags when rollbacks are impossible, and local rollback and automated deployments to shrink blast radius.
AI-written summary · about
chapters (13)
- 0:00Intro
- 0:12Why do we need software updates?
- 3:38Most Important Slide (Shownotes)
- 4:17Updating Faster
- 6:00Why do we update?
- 8:00Why is quality assurance hard?
- 11:43Tips to build trust
- 16:14Continuous over-the-air updates
- 19:00Updating servers
- 22:38Do Progressive Delivery
- 25:21Lessons: Don't do like Cloudflare
- 27:00Liquid Vision
- 27:56Corner Cases
related talks
50mDevOps Patterns & Antipatterns for Continuous Software Updates
17mDevOps Patterns & Antipatterns for Continuous Updates
27mDevOps Patterns and Antipatterns for Continuous Software Updates
48mDevOps Patterns and Antipatterns for Continuous Software Updates
05mDo this, not that: Antipatterns in Continuous Deployment
42mPatterns and Antipatterns of Continuous Deployments in DevOps Practice
46mContinuous delivery antipatterns
35mThe Dawn of Liquid Software
