I’m currently in the planning phase of setting-up my own Kubernetes cluster from scratch, with nodes around the world. 4 to be exact.
There is a difference on the Networks/Servers:
1x "Monster" 6TB Raid1, 20c 2.4Ghz, 256G Mem, 100Gbit net 2x "small" 25GB, 1c 2.4Ghz, 2G Mem, 10Gbit net 1x "storage" 20TB Raid1 (x2), 32c 4.5Ghz, 128G Mem, 1Gbit net
My original idea was to use the “storage” one as a backup of data and more or maybe even a replicate-able master, kinda like a control unit where the “Monster” would run the main applications and storage, and the small nodes would be application endpoints. but i’m unaware if this is even possible with kubernetes, and unfortunately I’m unable to find anything about it.
If i messed up my search please let me know..
Basically this is what i have lying around, can maybe switch the 2 “small” with something with more space, but possibly max 100GB.
Anyways, I was thinking of running something similar to GlusterFS across the “Monster” and the “storage”, for storage purpose for kubernetes, and then making “storage” the k8s-master, and the “Monster”/”Small” as nodes. Though this might be the wrong approach for Kubernetes.
Unfortunatly I have a hard time finding information on requirements of them master when we are talking about network etc. Of course more network is always better, but i was thinking of using the “Storage” as a backup for the others if all else should fail.
I would like some help honing in on what the best option is in this situation?