Change your tools, change your outcome
summary
Argues that deployment must move from an app‑only mental model to explicit system descriptions and orchestration, and demonstrates Bosch deployment by packaging GitLab HQ (sources and dependencies pulled from S3), compiling binaries, booting VMs, and scaling rescue workers live. Compares tool tradeoffs: Chef/Puppet provisioning vs declarative system manifests, JRuby benefits like JRuby introspection and thread choices, and logging/aggregation (Logstash/Cabana vs Splunk). Advocates app-centric vs system-centric thinking: describe full system, reduce external dependencies, and automate safe changes.
AI-written summary · about
chapters (43)
- 0:00Intro
- 1:03The Tshirt
- 2:11My background
- 3:08Rails consulting
- 7:10Fun challenge
- 7:48People suck
- 9:20Production app
- 9:40Constant traffic
- 10:17Magic move
- 11:12SOA
- 12:13Dashboard
- 12:46Basic Math
- 18:57What we should all care about
- 20:00Fundamental priorities
- 21:57Tools and consequences
- 22:27Every tool has some purpose
- 25:00Chef
- 25:40Handling errors
- 30:57Platformasaservice
- 32:08JRuby
- 33:33Visual VM
- 33:50Queues
- 34:13Logging
- 37:12Engine Yard
- 37:22Lab HQ
- 37:37Ghetto Light
- 37:49Touchscreen Monitor
- 40:33Release
- 40:58Manifest
- 41:52Parameters
- 42:09Deployment manifest
- 45:07Deploy manifest
- 46:15Add rescue
- 46:47Extra VM
- 47:17Auditing
- 47:27Rescue Workers
- 48:23Its just a process
- 48:54Save job descriptions
- 49:25Use shell scripts
- 49:40Load later
- 50:16Multiregion
- 50:57Summary of talk
- 51:43Hiring
related talks
27mDevOps and Baking
32mAdopting the right DevOps Tools for you and your team
01h 02mTools for simplified large scale systems
39mRunning Ruby Apps
57mGo Away Or I Will Replace You With A Very Little Shell Script
22m20+ years of building systems that outlast the chaos
50mExiting Vacuum: Integrating Configuration Management into your Ecosystem
25mDevOps for Pointy-Haired Bosses