Adding more disks to Storage Spaces Direct didn't give me the capacity increase I was expecting

Question

I have S2D setup using 2 Azure VMs in a mirror. When I originally set it up, I did so following the instructions here: https://docs.microsoft.com/en-us/windows-server/remote/remote-desktop-services/rds-storage-spaces-direct-deployment

This worked great with no problems. When I set it up, I chose to use 128GB datadisks. With 2 disks on each vm, I had 252GB of capacity in my volume.

Now I'm coming back to this set up and I wanted to add more capacity. I added a 1023GB datadisk to each vm. Waited for the storage pool to automatically add them, which it did. Then I tried to Resize my virtual disk using the information here: https://docs.microsoft.com/en-us/windows-server/storage/storage-spaces/resize-volumes

When looking at my Storage pool in the "Failover Cluster Manager" it indicated that I had 2.00 TB of free space. I assumed I could only use half of that, so my new capacity should be approximately 1.2 TB when combined with my original capacity.

When I ran the "Resize-VirtualDisk" powershell command it told me there wasn't enough available capacity. So then I started just reducing the size and finally got it to work if I resized to 500GB.

So adding 2 x 1TB disks only gave me an extra 250GB of capacity. Why is this? I've been trying to figure out and understand how this all works, but I can't find a definitive answer.

Answer 1

TLDR: try adding disks of the of the initial size.

---

It sounds like you are stranding your data. Storage Spaces Direct (S2D) is intelligent and can reclaim the data if with the right combination of extra nodes, for example, it has this sort of cleverness:

It acts playfully depending on its nodes/constraints. Simply adding a fourth node would seem to strand space, but actually allow for a mass reshuffling.

Related, you are creating reserve capacity (whatever you find unusable on your new 1 TiB disks). S2D has some extra intelligence here in that it will use this unclaimed space to reallocate bad sectors on the same disk(s).

Since you are doing this in the cloud though, I wouldn't worry about this stuff if you have read your SLA; just follow the docs and let Azure worry about bitrot. But if you are doing this on premises, it's definitely worth considering.