-
|
Hi Vito and all, at first thank you very much for this great project, I have used it for small demos and for learning, and always had a great user experience. I want to run a small "production" cluster for my side projects, and stumbled a bit on the recommendations regarding control plane location: In the Recommendations Doc Chapter Small to Medium Cluster 1-50 nodes, the small-to-medium cluster cp nodes are in one region (nbg1). In the same doc, in chapter Small Production Cluster 5-20 nodes, they are distributed across three regions. Personally, I would have assumed, having them in one region, would be better due to network latency vs. etcd's tolerance for such latency but I believe etcd can cope with the nbg/fsn <-> hel latency with some Tuning. The benefit of stretching is in higher resilience. Do you deploy the control plane in one region or do you stretch it? Thank you for your guidance and Happy New Year, Mitja |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments
-
|
Hola, I have been using a control plane across the three European regions for a while, but eventually decided to migrate the masters in Helsinki to one of the two German regions to reduce latency. It was working fine, but I kinda didn't want to risk to run into issues with etcd. So at the moment our clusters all have two masters in Falkenstein and one master in Nuremberg. Happy New Year to you too! |
Beta Was this translation helpful? Give feedback.
-
|
Just saw the notification of the sponsorship. Thanks! |
Beta Was this translation helpful? Give feedback.
Hola, I have been using a control plane across the three European regions for a while, but eventually decided to migrate the masters in Helsinki to one of the two German regions to reduce latency. It was working fine, but I kinda didn't want to risk to run into issues with etcd. So at the moment our clusters all have two masters in Falkenstein and one master in Nuremberg.
Happy New Year to you too!