"Change Management: A Scientific Classification"
summary
Classifies change management into a spectrum: manual live edits, heavy approval/sign-off, ITIL-style process, scripted automation, and full infrastructure as application with APIs and CI. Argues automation with Puppet and Chef plus virtualization/EC2 speeds deployment but tests often become technical debt, so continuous deployment is safe only where mistakes have low customer impact. Recommends roll forward not roll back, treating systems as disposable images rebuilt from code, and investing in tests and feedback loops so deployments actually reduce outages.
AI-written summary · about
chapters (24)
- 0:00Introduction
- 2:12The Dunning Kruger Effect
- 6:20Understanding Value
- 6:47Definitions
- 8:13ITIL
- 9:44The Six Laws
- 10:45Solutions
- 11:20Meat Cloud
- 12:22Signatures
- 13:41Master Plan
- 16:48Technical Debt
- 18:44What are we neglecting
- 20:35Continuous Integration
- 21:45Infrastructure as Application
- 22:18Proactive vs Planned
- 23:06People
- 23:59Developers and Operations
- 25:07Wall of Confusion
- 26:23Best Practices
- 28:20Use Your Judgement
- 28:55Be Proactive
- 30:07Roll Forward
- 30:47Specific Ideas
- 32:23Continuous Deployment
related talks
28mChange management (Keep it Simple, Stupid!)
07mAutomated Change Management
33mSystems Automation: Past, Present and Future
31mCouples Therapy for DevOps and ITIL
21mChange Management in Physical Network Lifecycle Automation
29mMutability, Declarability, and Disposability, oh my!
32mThe Necessity of Configuration and System Management Tools
01h 19mThe Business of Configuration Management