3 ms·
In 2000s we were talking that `snowflake` servers are bad. New generation is re-learning the same with k8s, which can be summarized as 'snowflake k8s clusters a
by betaby 3y ago
In 2000s we were talking that `snowflake` servers are bad.
New generation is re-learning the same with k8s, which can be summarized as 'snowflake k8s clusters are bad'.
Fundamentally it's the same problem.
- menschmanfred 3y agoIt's not. The control plane is ha and you can upgrade one after the other independent of your workers. With workers you can do that too. That feels much less like a snowflake and more like snow.
- betaby 3y agoControl panel as HA as your certificates which were expired in the article.
- menschmanfred 3y agoK8s introduced Auto rotation surprisingly late. Even we run into that issue 5 years ago. But k8s is still very young and the problem is solved for 5-6 years
- chupasaurus 3y agoIt was introduced later than the related crash in the article.
- menschmanfred 3y agoYes. But the article describes years of usage and it just isn't a problem if you start with k8s today
- dilyevsky 3y agoEven if control plane is hard down your kubelet wont be evicting anything so unless you need to surge or everything crashes you have some time to fix things. I’ve dealt with multi-hour cp outages and while stressful it didn’t have any customer visible impact whatsoever
- dilyevsky 3y agoare snowflake products bad too?