[PaulHoule]

It was developed to solve Google’s problems and not developed to solve the problem of other organizations. It was better than the half-baked attempts Docker made at container orchestration and for whatever reason no real competitor has emerged. It doesn’t have the usual forces acting on it that cause products to bend to the needs of their (ex. Google) users.

I like this product vision

https://en.wikipedia.org/wiki/IBM_Parallel_Sysplex

which takes a very opinionated approach to the problem of distributed systems circa 1995, particularly in that it revolves around centralized timekeeping and synchronization services that are high performance for “small” clusters (two racks worth of machines) but not scalable to the absurdly large clusters that Google, Facebook and a handful of other companies need. It adds up to something that is refreshingly manageable even if the documentation is full of idiosyncratic vocabulary.