Struggling to get this conceptually, so hopefully someone here can shed some light on this.
We're trying to implement storage spaces on a single 2019 server having (2) 800GB SSD's and (5) 1.8TB 10K SAS drives. The goal was to set up a mirror of the SSD's and use it to accelerate a RAID5 of the spinning disks (we have one hot spare). It seems that when you create a pool of all disks listed, you can only mirror to get parity (I'm assuming because we have only 2 SSD), so I think I'm approaching this wrong (thinking it's more like a SAN based performance tier which can be on different raid types than the standard tier). Should we be adding a third SSD so that we have uneven numbers of both disk types or am I going about this all wrong? I based the theory on a Storage Spaces Direct article which I'm now thinking is not applicable in a single server deployment (https://docs.microsoft.com/en-us/windows-server/storage/refs/mirror-accelerated-parity).
When we use all of the disks in a mirror pool (burning more spinning disks than desired), we then can't deploy a single vdisk across the pool using the maximum space (I'm assuming because the uneven disks on the spinning side creates an issue with virtually writing even mirror pairs). By simply splitting it into 2 volumes, it seems to work much better and allows us to use almost the entire pool (hard to get the vdisk sizes perfect to get an even split). We basically want one pool, tiered to house Hyper-V guests. Some of them will be Hyper-V replicated to another hosts set up the exact same way. They're in a separate IP subnet, so replicas are the simplest way for us to handle a failover (as opposed to a cluster).
Any thoughts on this topic? Searching has been tough as everything references Storage Spaces Direct and my eyes are blurry from trying to separate the two methods.
Thanks in advance,
Eric
