Re:Docker in Production: A retort
by empath75 on Saturday November 05, 2016 @06:34PM
(#12882298)
I think old school sysadmins care way more about system stability and new school devops guys are focused on cost and the speed of the development cycle. It really doesn't matter if you architect your systems properly that docker is unstable or your services crash as long as you have effective monitoring and self healing systems.
[ Parent
| on HN ]
I'm both or rather I do things like this:
* Use modern tools to spin up a dev env * Determine how to monitor and maintain dev and deploy * Test dev env * Determine, based on dev and a deep risk assessment how prod will look * Create prod env * Determine how to monitor and maintain prod env and deploy (based on your test) * Test new prod env * Open up prod env to end users
It's not bloody rocket science - it's Engineering and the steps above hide loads of steps and feedback loops such as a safe upgrade method eg snapshot/clone VM, test, release.
If your product is important to you then you take it seriously and treat it right!
Cheers Jon
PS Did I mention documentation?