About a year ago I accepted a new job in an enterprise Linux environment, running ~450 Linux servers. These servers were running on an internal network and had never been updated. Most work was done ad-hoc and in response to issues or failure.

I transformed the team to a pro-active way of working where automation was key. By solving the most frequent problems first, we found the time to automate more and more. Every server was updated and configuration management was introduced.

One interesting year later I've automated myself out of my job. The team can easily handle the (now much lighter) workload without me. Users are happy, so mission completed!

Configuration management is done using CFEngine 3 and we use other DevOps style tools like Git, Vagrant, Trello, Logstash etc.

In this talk I'll talk about the journey we took and answer:

• What was the problem with the old way of working?
• How did we implement configuration management in an already running environment?
• Why did we choose CFEngine 3?
• What is the new way of working?
• What are the lessons learned?
• How to automate yourself out of your job?

To get an idea: I have recently written blog post about the technical side of the story: http://blog.remibergsma.com/2013/08/31/managing-hundreds-of-servers-with-cfengine-and-git-full-control-over-configuration/

The presentation will also focus on the process of getting there and the lessons learned, so others can benefit.