[brutus1213]

Totally agree with your first paragraph. Regarding execution, I'm not sure what Docker could have done differently that did not lead to the outcome we have today. I don't think that playing nice with other opensource devs would have made a difference (as the article claims).

Also .. the K was hardened at Google is BS. The ideas, maybe. But I am quite skeptical about the amount of prod internal google code went into early K (pls don't point at the Borg paper .. I'm talking about actual working code). I recall doing a deep comparison of swarm vs K circa 2015 and Swarm was clearly superior in both design and implementation. Today, K is better and has an ecosystem .. maybe the issue isn't that core Docker containers played nice with opensource .. rather .. swarm should have focused much more on playing well with others.

One point of contrast is Hashicorp .. they are in the workload orchestration and mgmt space and seem to be doing really well. Kudos to them!

[smarterclayton]

Kubernetes (from open sourcing to about 1.3 or 1.4) is a second system mostly written by senior engineers (from several companies) with deep experience in the problem domain and strong architectural guidance, and a willingness to stop at “just good enough” and then let stuff mature. Kube was mostly “done” from a design perspective in early 2015.

Swarm was 2-3 people in the early days, without as much strong opinionation about what exactly they were building, which meant while it was a tighter, simpler system, it couldn’t evolve as easily.

I’m obviously biased - I was the first non googler to have commit on the repo. But it’s much easier to build something when you know upfront exactly what it looks like and you have a set of committed and experienced engineers with good leadership.

[SPascareli13]

> I’m obviously biased - I was the first non googler to have commit on the repo.

How well was your PR received?

[forgot-my-pw]

Looks to be successful in general: https://github.com/kubernetes/kubernetes/pulls?q=is%3Apr+aut...