Super interesting response to my thread yesterday from @jmoiron (h/t @copyconstruct). Thank you! There is a lot to digest there and I don't want to do that thing where I deconstruct it point by point so I'll try to hit some of the high points. jmoiron.net/blog/is-k8s-to…
First off: Jason is frustrated and his experiences are valid! This stuff isn't where we need it to be and there are undoubtably things that won't pan out over the long term. We are still early in this game and it is chaotic and exciting and confusing. I share that frustration.
Next wrt marketing: we are at the height of the hype cycle for Kubernetes. There nuggets of goodness can get lost in the effort to thought leader in more and more grand themes. Marketing gets out of control. This really bugs me and makes me grumpy.
For me this was exemplified by the breathless and commercial nature of the keynotes at KubeCon EU. I loved the hallway track and the breakout sessions but there was a over-commercialization tone that I didn't like.
I also want to explicitly call myself out on my own BS. This tweet () is about how engineers are blind to their own complexity. That applies to me. I'm not the right person to judge if k8s got things right nor not. Users like Jason _are_. I'm listening.
Two concrete technical things were brought up in the post as critical pain points: YAML and k8s discovery being wrong. I 100% agree with YAML and I need more data to understand the issues around discovery.
As @bryanl says: YAML is for computers. When we started with YAML we never intended it to be the user facing solution. We saw it as "assembly code". I'm horrified that we are still interacting with it directly. That is a failure.
This is not an easy problem and I don't think there is a silver bullet. By having a raw form we do enable an ecosystem of solutions. Those are still early and are a part of that raw chaotic primordial soup that is parts of Kubernetes at its current stage.
From @heptio's point of view, we've been investing in @ksonnetio. Bryan gave a great talk at KubeCon on this and is taking a *ton* of time to help understand where the pain points here at various scale levels.
We run the risk of solving problems by introducing even *more* complexity. I worry about this with efforts like Istio. It can do amazing things but is both early and yet more complexity piled on top. We have a limited capacity to absorb this stuff and k8s isn't digested yet.
Moving on to Service Discovery -- I suspect the object is about either labels or the fact that the system registers endpoints vs self registration. There are reasons and scenarios for both of those. But I won't monologue without more info.
I wrote a lot about the how and why of service discovery in Kubernetes in the Kubernetes Up and Running book. That chapter is included in the excerpt we are hosting: go.heptio.com/kubernetes-up-…. (Warning: email capture form there).
Summary: I'm pretty sure we got core concepts mostly right in Kubernetes but I'm not the one to judge. It *was* based on lots of experience at Google but that experience doesn't always translate. K8s isn't perfect but it also isn't finished. Ecosystem is chaotic and exciting.
