Easy to configure ?

 

A few years ago, in the early days of Kubernetes all was fine.

I came with over 15 years of developer experience in Java EE  and also knowledge in infrastructure topics as well as hardware, and I started my cloud journey with Goggle Cloud and RedHat Openhshift.

Although I already did development in CloudFoundry, learning Docker, Containers and the all the Kubernetes Resources was tough.

I'm always learning trying to understand the "Why" behind the "How". Finally I thought I know all the about the Kubernetes Resources and their meaning.

Ok, maybe Kubernetes API will grow, but that would be manageable.

...

Then Kubernetes Operators came into live, and I really liked the idea of using operation teams knowledge as a recipe for automatic reconcellation loops.

Over the time the amount of those operators in their catalog repositories, grow and grow. Some were community based, some were maintained by company.

They serve diffent purposes from the CNCF landscape: databases, monitoring tools, service mesh.

But almost every operator came with the promise that it would be "easy to configure".

....

To be honest I doesn't think they are 😒

I always had to invest a lot of time in learning CRDs  and their params.  I learned that "kubectl explain" is my friend in doing that.

Some operators came in addition with CLI tools to ease configuration, but at least they always produce CRDs as well.

As that is not worse enough I sometimes found out, that not everything was configureable via CRDs and I sometimes end up in learning the configuration of the underlying tool as well.

In real world enterprise use cases, i never found it "easy to configure". 

So let me warn you: 

"easy to configure" is the same as "it worked on my machine"