DevOps should boost delivery and stability. Without a clear roadmap and strategic deployment of DevOps tools though, your organization won't enable continuous delivery.
I have a secret for you. You know the "DevOps specialist" in continuous integration your company is trying to hire?
That person does not exist. Or, at least, the few DevOps specialists that really do have five years of experience will cost more than your employer will want to spend. No, your company is a part of the late majority. The word "majority" here means there will be a sudden influx of demand for DevOps before there's a true capacity.
That is where you come in. You are going to fill the gap.
DevOps is tool-obsessed
The place to start is the continuous integration (CI) pipeline -- the one that connects checked-in code to a build. Over twenty years ago, Steve McConnell called it “"daily build," "the heartbeat of the project." Your goal is to speed up the build, while reducing risk and enabling faster deploys, rollbacks and monitoring.
A great deal of DevOps literature is tool-obsessed.
That's where DevOps starts.
A great deal of DevOps literature is tool-obsessed. We talk about virtualized images created by code and deployed in the cloud over a cluster manager, and so on. These are big words. Infrastructure as Code, feature flags, blue-green deploy tools -- all of those are implementation details. The important thing is what they do for us.
Developers, system administration and testing need to work in unison.
Start with the Heusser Test
So here's a way to start. In the next Agile retrospective, identify a way to improve the build system that will also boost performance -- and offer to do it yourself over the next sprint. Suddenly, you are the DevOps team. Keep it up for a year and you will become one of those "DevOps specialists" in CI I was talking about in the beginning.
"The Heusser Test" looks at this idea with a little more depth and details "six steps to better code faster." My goal was to come up with a sort of checklist that gets developers, system administration, and testing to work in unison and therefore cut time-to-market and improve quality. It also walks you through building out that CI pipeline into a true continuous delivery (CD) pipeline with rollback.
One last thing. A few months ago, I worked with a Fortune 500 software firm moving to CD. I met the entire company's CD team, who were building the infrastructure. I asked where they would be in 18 months. They had no real answer. I asked for the roadmap; they didn't have one.
I asked a third time and finally heard back.
"It depends on what the teams do, but we don't know for sure. I mean, we aren't going to be putting WebSphere in a Docker instance."
