In the course of my job I see hundreds of DevOps setups a year (340 in 2020 to be precise). GitOps is everywhere. One of those hyped concepts everybody has to do. I started realising that all the teams I talked to roughly split up in two teams: “hate” and “love” GitOps. The “love” team is much larger. With growing sample size I started mapping the satisfaction level towards GitOps against the size of installations (as it turns out against the amount of YAML files).
That was the point I started to look deeper into the topic, trying to understand the “whys”. I was inspired initially by Adam Sandors article (GitOps the bad and the ugly) that triggered an intense discussion on here. I summarized my findings in an article I call “The Fundamental Flaws of GitOps – A Statistical Analysis“. My finding is that you can explain the general sentiment towards sentiment using the equation: P(GitOpsfail)=1-p^n; Sounds strange but I do believe it has a good amount of truth to it.
Don’t get me wrong: GitOps is a great concept if you’re running small amounts of services – I do not believe it scales for the enterprise.
To balance this we are organizing a webinar with Viktor Farcic (Codefresh – GitOps tool), Adam Sandor (Containersolutions) and Chris Stephenson (Humanitec). Feel free to join and actively participate in the discussion.