I'm going to setup a VM, starting with two disks, one for / and one for the data using a ZFS-pool. Don't want to create VM disks too large, even in case of dynamically growing ones, and prefer chunks of e.g. 250 GiB per disk. Makes backup, file handling etc. easier in various cases in my opinion. So at some point the VM might end up having 4 virtual disks instead of one large one with 1 TiB. The only downside of that approach is that ZFS thinks to have 4 individual disks and spans I/O and data over all of those 4 disks.
Is there any way to prevent that and make ZFS use a JBOD-like approach? So that while it does use all of the disks, e.g. it simply writes from the start to the back in a linear way?
I'm aware of possible workarounds like LVM within the VM, but that's unnecessary difficult. I can easily live with the downside of the ZFS-setup and did so in the past, wasn't that much of a problem even when the virtual disks were backed by HDDs only.
I didn't find anything and was wondering if I missed something or it's simply not supported.