There’s No Nice Way to Say This: Your DevOps Has Gone Horribly Wrong

Editor’s Note: A popular session at All Day DevOps, 2019 inspired this post. This Friday, April 17, All DevOps | Spring Break debuts with brand new speakers and perspectives from around the globe. There is still time to sign up and join the largest group of DevSecOps professionals making the best of being indoors. Join us! 

DevOps is not about the tools. Nor is it about the number of people working on any given thing. Adding people doesn’t mean delivering faster. You’ve got to measure what matters. It’s about principles and culture! Kalle Sirkesalo (@Failattu) shares his perspective.

Sources of Problems

Using DevOps practices can make it easy to add environments as the team grows. However, each environment adds to the pool of technical debt. All of the tools can be in place and you can have proper Gitflow branches, CI/CD, and automated testing. Even though everything is technically correct, organizations can have problems delivering.

What both John and Kalle are getting at here is that you need more than just the cool tools to succeed. And, in some cases, the tools can compound existing problems, making things worse instead of better.

Long-Running Branches

One source of trouble—a practice smell if you will—is long-running branches. These branches become difficult to merge and difficult to test. But this just a symptom of a deeper problem. If something like having too many environments is blocking your deployment, you may end up with long-running branches. Merging your changes into a branch for each of these environments invites difficult merges. You’ll have to merge over and over again—once per environment branch.

Different Processes

Another compounding factor of making life difficult for developers, ops, and the business alike is having teams that operate under different processes. Misaligned hierarchies can create political (Read more...)