The Deployment Manifesto

heystephenwood.com

2 pointsfaddai2 comments

Can you explain rule #3?

"Devops SHOULD work with engineering teams to come up with continuous deployment strategy that doesn’t involve the destruction and creation of fresh operating systems."

Why should one avoid a deploy process that destroys and creates an OS? Can you give an example?

I'm using MS Azure and one of the PaaS solutions is a "Cloud Service", where you give them your code and a definition file that describes what the OS should look like. They then provision you a fresh VM. Takes about 5 minutes to do a deployment. A bit time consuming but overall I find the process really convenient.

And I find that this means the entire build out of the web server is scripted. Meaning at any time we always know what settings are involved, and what 3rd party dependencies should be setup and how they should be configured, ACL's, etc.

This also makes scaling the # of instances up/down literally as easy as updating a single line in a config script, or adjusting a slider in the admin portal.

Reply on news.ycombinator.com