We’re in the process of performing a large physical-layer upgrade with our hypervisors and I am looking for some advice regarding our proposed configuration and VSAN Software. To break things down, here’s a high-level view of our current working infrastructure. All servers are using the Hyper-V virtualization platform and we’re planning on sticking with that.
· HV01 (production / old cluster)
· HV02 (production / old cluster)
· HV03 (Extra server / un-used)
Right now, HV01/HV02 are built into an active/passive cluster utilizing Veeam’s replication. HV03 is an extra server we have on-hand from another project. For reasons outside the scope of this post I’ll deviate from going into the details of why this is not working out for us. We are the new IT consultant for this company, and this was the way the network was originally configured. The working plan is to move all VMs to a single node (HV01), and then combine HV02 and HV03 into the new production cluster. That’ll look something like this:
· HV01 (old cluster / All virtual machines)
· HV02 (new cluster / pending VM migration)
· HV03 (new cluster / pending VM migration)
Then, we will be upgrading all of our Server 2008 R2 VMs to newer editions of Windows and migrating them to the new cluster. Once the upgrades and migrations are complete, we will roll HV01 into the new cluster to have a total of 3 nodes.
· HV01 (production / new cluster)
· HV02 (production / new cluster)
· HV03 (production / new cluster)
All servers are Dell R510s with 12 drives on the front back plate. All 12 drives are sitting in front of a PERC H700 RAID controller. Because we have so much local storage, we’re looking to visualize our SAN and potentially cut the new hardware out of our bill. This is where we’re running into some problems – the software. From many recommendations on this forum, Starwind’s vSAN seems to be the way to go and that was my first inquiry. Unfortunately, their cost is on the higher-end of our budget, and if possible, we’d like some other recommendations the community may have. S2D is out because of the PERC RAID controller. If push comes to shove, we’ll go with Starwind’s, I would just like to see if any other options are available.
This brings me to my next question. Due to some bad-timing, management wants to add another VM to the current production cluster. This conflicts with the original plan of migrating all production virtual machines to HV01, we’re barely scraping by on available resources. Needless to say, we wouldn’t have room to add any additional VMs.
Using Starwind’s (if that’s what we go with) would it be possible to create a new, 1-node, fail-over cluster out of HV03? Then as we progress on our regularly scheduled time-table, we can add HV02 into the cluster and replicate local storage (with the new VM).
Apologies for the wall of text, and convoluted nature of the question, but any assistance and recommendations would be much apricated.