Hello all!
I have vSAN stretch cluster 12 nodes. Have one disk group per node (All flash).
1 x 1.46 TB Cache
4 x 3.49 TB Data
What is the best practice for expanding. I was thinking one more Cache disk and two more data disks so it would be:
2 disk goups per node:
1 x 1.46 TB Cache
3 x 3.49 TB Data
+
1 x 1.46 TB Cache
3 x 3.49 TB Data
1. These disks are expensive, could just have 1 more Cache and one Data in the same disk group, so it would be:
2 x 1.46 TB Cache
5 x 3.49 TB Data
Please advice 🙂
Hi
I suggest you have more disk groups per node, to have greater redundancy in case of a disk failure. If you use only one diskgroup in the event of a disk failure you would lose the diskgroup of that node and therefore all the space of the node. If, on the other hand, you use more diskgroups per node, the failure of a disk only impacts on a diskgroup and therefore only on part of the node's disks. You also have performance gains.
These are somewhat old but interesting links:
vSAN Failure Scenarios - Virtual Blocks Blog (vmware.com)
One versus multiple VSAN disk groups per host | Yellow Bricks (yellow-bricks.com)
Yes i agree that a second disk group per host would be the best option to increase performance and redundancy. You can only have 1 Cache disk per disk group so your option suggesting to have 2 cache per disk group would not work.
Like others have mentioned it's not possible to have more than 1 Cache disk per disk group. I have the same advice as the others, pleas do more than 1 diskgroup.
So in all that would mean for you (per ESXi host):
DiskGroup 1:
1 x 1.46 TB Cache
2 x 3.49 TB Data
DiskGroup 2:
1 x 1.46 TB Cache
2 x 3.49 TB Data
This way you would potentially only need 1 Cachen drive extra per host.