Software fault-tolerant storage?

Question

Storage Area Network

Good day to all!
There are: several high-performance servers of the same type with OS virtualization functions (so far on Proxmox), a bunch of other various servers, switches (optics and copper), direct hands.
Task: Build a failover virtualization cluster with failover storage.
Question: what product (software) to choose for building fault-tolerant storage?
Requirements: minimum downtime (i.e. one node broke, the second picked up) and "guy" settings.
Пока смотрю в сторону Старвинда (но хочется, конечно, чтобы не на винде оно всё вертелось). Читал про Ceph, но по комментариям не понял, подходит ли оно для хранилища, построенного на двух-трех нодах, потому что пишут, что умирают диски после перебалансировки нескольких террабайт.
Рассмотрю любые варианты (не только коммерческие).

Answer 1

CEPH works for us, disks do not die during rebalancing! The storage itself is now 119TB on 21 OSDs in 5 servers.
The flight has been normal for 3 years.
For ceph, at least 3 nodes are desirable. The storage itself is very convenient and flexible.
If you want to get good speed, then for interconnect inside ceph take ethernet 10Gb (we have 20/40Gb infiniband) cards and 1/10Gb cards for clients.
In principle, it takes 15 minutes to raise a new ceph cluster, but you need to be careful during installation.
Ask questions, I will answer.

Answer 2

+1 vote for Starwind. Under windows, I only know storage spaces from win2016, but I need 4 servers with an enterprise license. Expensive. Factor of.
In general, I work with starwind: 2 physical. server, hyper-v cluster, everything is simple and transparent. There is a third server with hyper-v outside the cluster with a VM with an additional domain controller and other non-critical ones.
Starwind has Russian-speaking technical support again.