I wanted to get into a bit of a exploration of the reasoning behind the virtualization of an unRAID server and see what kind of use cases fit into this mix well, and which ones fall outside a good use.

 

I have a simple ecosystem at home that looks like this. An Unraid on a SC846 (24 bay LFF 3.5) and proxmox on a T620 (32 bay SFF 2.5). Simple GBE and 10gb links without a fancy router like Sophos XG/PFsense (this however is on the project to-do list asap) and a UPS on each. 

 

The first use case I am interested in the nice live migration features of Proxmox for VM's.

 

One machine is much more useful for loading up SSD's and running VM's (t620) and one is geared more toward being a large filer (SC846). Indeed this is how I have them running now. I have ~15 VM's running on the t620 and currently use the 846 for Docker and have ~8 running on the SSD cache. 

 

Due to the need to passthru the host flag on proxmox to run unraid and the fact that I cannot move drives with ease, this seems to be one-way migration off the t620 and onto the sc846. Both machines have performed very well and not needed resets so I am just not sure this is a great use case fit.

 

I do have a backup system for both, a tape library, so I do not consider either of these machines to have that functionality. 

 

In writing this I think I have talked myself out of it.

 

i have a very similar ecosystem with two servers - one for like a storage, and one for like a VMs, both connected with 10Gbit network.

i have proxmox on both on them, in one cluster. unraid is running on storage one (2VMs), most of the VMs is running on second one. on both servers i have setup NFS servers, and they serves storage to each other. with this setup i can easily achieve VM migration on fly (except unraid VMs of course), but other ones works as expected. i now, NFS layer adds some overhead but for my use-case it works just fine.