Choosing a server for a high availability cluster?

Question

Hyper-V

Good afternoon!
The situation is as follows:
There is a server with the hyper-v role. Raised 6 virtual machines, which are located on it. There are no performance issues.
There are also two NAS: QNAP TS-412U and Synology RS812.
On the first one, data backups are made from virtual machines by Acronis, the second one is idle for the time being.
Server specifications:
Intel Xeon E5-2630 v2, 2533 MHz (single processor)
Motherboard: Supermicro X9DRW
Chipset: Intel Patsburg C600/X79, Intel Ivy Bridge-EP
RAM: 32 Gb
HDD: RAID 10 of 4 SATA HDDs 1 Tb
each Planned guides organizing a high availability cluster. We need to offer different solutions.
As for the budget option, I think to offer the following:
CPU Intel® Xeon® 10C/20T E5-2630v4 2.2-3.1GHz/25M 85W x 1
RAM 16GB DDR4-2400 ECC Reg x 2
RAID controller Adaptec RAID 8405 (4port SAS3, 1GB) x 1
Ethernet controller, Quad port, Intel x 1
HDD: 600GB, 2.5", 10000rpm, SAS x2
Price: ~201,000 rubles.
That is, I plan to transfer virtual machines to a LUN that will be created on one of the existing NAS and will be connected via iSCSI as a shared disk space to the cluster nodes.
Backups I plan to make virtual machines on the second NAS
Questions are as follows:
1. Will the cluster work from a server on the Supermicro platform and a server on the platform, for example, IBM Or should the servers be from only one vendor?
2. Network equipment. As I understand it, there should be a network of two nodes, storage with shared disk space and a NAS for backups. What network equipment would you recommend for networking these nodes?
3. In the event of a failure of the storage where the virtual machines are located (it is planned to deploy this on a NAS), should the other NAS where the backup copies of these machines are located then become a shared disk space, or is it better to do it right away and connect to the nodes via iSCSI? The issue is to ensure that there is no long downtime in the event of a failure of the storage that hosts the hard disks of the virtual machines.

Answer 1

1) Servers can be of any vendor, but the models and generations of processors must match. In your case, this is not observed. VMware has EVC technology, which allows you to equalize the instruction sets of processors of different generations by the oldest of them. As far as I remember, Hyper-V has nothing like that. Therefore, if you have processors of different generations and models in your cluster, how not to rake problems, so smoke this question.
2) If we are going to walk for full high availability, then there should be at least two switches, and which ones specifically - it depends on the loads. Maybe 1G is enough, or maybe some kind of InfiniBand is needed :-) Two switches, each of which is connected to one interface from each server and each NAS. So that if any of the components fails - a network card, a patch cord, a port on the switch, the switch as a whole, the connection is not interrupted. But this will, of course, require iSCSI multipath and/or interface bonding.
3) In your case, it is better to connect both storages to hosts at once. And keep backups somewhere else. To reduce downtime for restoring virtual machines from backups, you can use the mechanism of replication from one storage to another. For example, Veeam can do this. Acronis, I believe, too, and there are some built-in tools in Windows on this subject, you just need to figure it out. Then the downtime will be minimal, and data loss is negligible.

Answer 2

I'm scared for your availability, if you ask for advice for it on the Toaster.