Just chuck in my two cents worth.
I have a three node Proxmox cluster.
In each server we have a hardware raid controller and also an LSI 9211-8i HBA. The first 4 ports are on the HW RAID the second 4 ports are on the HBA
I have 8 disks in each. 2x120GB SSD on the HW RAID, 2 x 146GB 15K SAS on the HW RAID and 4 x write intensive SSD on the HBA
The Proxmox OS is on the two 120GB in RAID1, same with the 2 x 146 for just temp files and the 4 SSDs are CEPH
I have multiple 10GB links between in box direct with DAC cables in a broadcast bond, ie no 10GB switch needed.
The 10GB carry CEPH and cluster traffic.
We have 1GB feeds from individual switches of the datacenter into each of the Proxmox boxes.
It has never let us down, even when a 10GB card failed it just plodded on. I have also upgraded it no problem.
I have a three node Proxmox cluster.
In each server we have a hardware raid controller and also an LSI 9211-8i HBA. The first 4 ports are on the HW RAID the second 4 ports are on the HBA
I have 8 disks in each. 2x120GB SSD on the HW RAID, 2 x 146GB 15K SAS on the HW RAID and 4 x write intensive SSD on the HBA
The Proxmox OS is on the two 120GB in RAID1, same with the 2 x 146 for just temp files and the 4 SSDs are CEPH
I have multiple 10GB links between in box direct with DAC cables in a broadcast bond, ie no 10GB switch needed.
The 10GB carry CEPH and cluster traffic.
We have 1GB feeds from individual switches of the datacenter into each of the Proxmox boxes.
It has never let us down, even when a 10GB card failed it just plodded on. I have also upgraded it no problem.