Kris Buytaert and I presented together at T-DOSE 2010, telling the story of why devs and ops needed to stop fighting. Kris started with the familiar scenario: developers writing code, throwing it over the wall when the deadline hits, and operations scrambling to deploy something they know nothing about. The devs think ops is blocking them. Ops thinks the devs have no idea how to run anything in production. Business gets fed up, fires everyone, and the cycle repeats.
The answer we proposed was infrastructure as code. With virtualization and cloud, we went from managing ten machines to managing a thousand, which meant automation was no longer optional. Configuration management tools like Puppet let you define what you want your infrastructure to look like in code – version-controlled, reproducible, and testable. You can throw a machine out the window and be back up because everything is code, not someone’s wiki notes.
From the development side, I explained how agile practices apply directly to operations. Backlogs, planning poker for estimating work, iterations with working deliverables every two weeks – none of this is exclusive to writing application code. If you are provisioning servers, deploying software, or managing infrastructure, these practices work just as well. The key shift is from solo heroes doing manual work to teams collaborating on shared, automated, tested infrastructure.
The real message was that getting devs and ops to drink beer and eat sushi together is a great start, but the structural change comes from sharing a codebase, sharing a backlog, and sharing responsibility for the whole delivery chain from code to production.
Watch on YouTube – available on the jedi4ever channel
This summary was generated using AI based on the auto-generated transcript.