3 Crazy Ideas to Make your DevOps Proposal Seem Reasonable
I work with “The Enterprise” everyday. Sadly, this has nothing to do with Star Trek – just big companies with lots of internal politics. As our friends champion our products internally,the political challenges are often tougher than the technical ones.
If politics is going to be a barrier, it’s fair game to use political theory to fight back.
The Overton Window
The Overton Window is a concept that has been gaining traction over
the past few years. The idea is that people are scared of extreme ideas,
and prefer moderate ideas. To push an idea towards the idea of
acceptability one can work hard to persuade the group that it is
reasonable. Alternatively, an ally propose a new and more radical idea
that makes your idea seem reasonable by comparison.
For example, in countries where spending money to reduce greenhouse gasses is controversial, a cynical party that supports the idea might find that proposing Geo-Engineering solutions could be more effective than arguing the merits of the idea. The “crazy” idea of Geo-Engineering would make reducing emissions seem practical.
So if politics is slowing you down, consider a two-pronged approach. Make rational and coherent arguments in favor of DevOps ideas while also manipulating our corporate Overton Windows by circulating more radical ideas.
For starters, here are some of my favorite radical ideas.
1. Continuous Deployment
Trying to move towards regular deployments to test environments? Or want to push the production release cycles to monthly rather than semi-annually?
Pass along the Timothy Fitz classic “Doing the impossible 50 times a day” which outlines how his company deployed their core money-making application to production dozens of times per day, every day.
2. Netflix’s Simian Army (of doom)
Many companies are lucky to test their disaster recovery capabilities once a year, and it’s a painful process. A friend of mine described how his team switched to failing over twice a month, every month. It was painful at first, but quickly became a non-issue and his team had confidence that in a real disaster, they would fail-over extremely gracefully. The self-induced pains related to starting to regularly test failover can look hard hard to accept.
However, Netflix is several steps further on the crazy scale. They created a collection of processes than intentionally take portions their core application offline with no warning. By comparison testing DR regularly looks like a day at the beach.
Share Netflix’s Simian Army around the office and see who’s inspired.
3. Just say NoOps
The “NoOps” gets almost as much attention in the trade press / blogosphere / twitterverse as DevOps. While similar to DevOps, it seems less pervasive in the real world. It is however, much scarier sounding to Operations people than working closely with Developers.
If a colleague would point out resources on NoOps your DevOps
suggestions will look much more friendly. Some of the operations teams
that I know who work most closely with developers started to do so only
after the dev teams started looking to EC2 as a solution to their
problem of too few test environments.
This feels manipulative…
Talking in terms of Overton Windows is manipulative and cynical. I get it. I’d encourage you to pay attention to the cutting edge nonetheless. When your proposal is being treated as extreme and crazy, it can be helpful to point out that real companies are very successful doing things that are far more extreme. Everyone should know that what you are proposing is, while a significant change, not the idea of a crazy extremist. The crazy extremists are doing far cooler things.
With that in mind, we’ve included an “Extreme” category in our Continuous Delivery maturity model wall chart and white paper. The white paper in particular covers some other great “extreme” techniques in build and deployment alongside the more pedestrian.
* Overton Window image from wikipedia.