I have a small oVirt cluster running but I have been having problems with the storage and would like to completely start again with the storage infrastructure. Currently for storage I have a single server running windows storage server serving oVirt via iSCSI.  I also have a smaller storage server which is currently not used. The oVirt cluster is not running a production environment but ideally I do not want to have to power it all off to patch the storage servers and oVirt cluster.

1x storage server, raid 10 24TB usable, 2x 10Gb ethernet

1x storage server, raid 10 4TB usable, 4x 1Gb ethernet

8x oVirt hosts, 1x10Gb ethernet on each

1x 24port 10Gb switch

What would be the best way to utilise the storage servers? 

I was thinking about sticking CentOS on both servers and running Gluster with a 4TB replicated volume across both servers for the hosted engine and other critical VMs and then a 20TB non-replicated Gluster volume running on just the larger storage server for non critical VMs. I have another spare server which i could potentially use as a arbiter node. Would this work or would I have huge problems as the hardware performance of each storage server is so different?

Any advice appreciated.