I currently have all of my HDD storage combined as one mount (44TB), and all the SSD storage combined as another (2TB). I put anything that really needs speed, like VM disks, Docker volumes, nfs-exported home directories, etc on the SSDs, and all ISOs, images, backups, etc on the HDDs. In a sense I'm micro managing the placement of data.
I'm about to move everything into a hyperconverged 4-node proxmox cluster, and I'm wondering if I should keep doing the same. The alternative would be to have a single storage and use the SSDs as cache in front of the HDDs. The upside is I don't have to micromanage (which isn't really that bad), but the potential downside is relying on the cache implementation, whether that's ceph, LVM, zfs, etc. Not sure how well it will work.
Any thoughts?
I'm about to move everything into a hyperconverged 4-node proxmox cluster, and I'm wondering if I should keep doing the same. The alternative would be to have a single storage and use the SSDs as cache in front of the HDDs. The upside is I don't have to micromanage (which isn't really that bad), but the potential downside is relying on the cache implementation, whether that's ceph, LVM, zfs, etc. Not sure how well it will work.
Any thoughts?