Is it possible to combine servers with NAS4FREE into 1 shared disk?

Question

linux

Colleagues, good afternoon
Please tell me if it is possible to combine several nas4free servers into one common disk pool, as well as further expansion of space due to additional servers
or tell me systems that can do this

Answer 1

And what is the purpose?
In general, it all comes down to GlusterFS vs Ceph. From reading tyts and tyts
In general, there are somehow more guides on Ceph

Answer 2

For example GlusterFS or DFS. Read about distributed file systems.

Answer 3

In theory... YES. You mount network drives....
If the host machine will support e.g. iSCSI, SMB, NFS protocol... but

to one shared disk pool

Each nas4free will be a separate folder (not sure if you can mount everything in one folder).
If you want distributed storage - then you've got it)))
I don't have it at hand ... I don't remember what it's called ... proxmox can work with such distributed file systems.
at least 4(3) computers, connecting other computers via the command line.
Played with technology on old system units. got a maximum read / write speed of 6 Mb / s, dropped to 1 Mb / s. In short, the speed is bad. Solutions for such systems: stable network speed of 1Gig and sufficiently powerful processors. Somewhere ... Someone even releases system engineers just for this ... It cost in ancient times for 1 million rubles.
Question - does it make sense? )))

Answer 4

Google about distributed file storage. But it will not be fast - surely you want to unite the most Dishman pieces of iron? As far as I understand the idea - there is a certain "center" that gives "outside" the representation of the "single storage", and already he monitors which blocks to write where. So, in order for all this to work at a reasonable speed on the NAS, in addition to the axis, the program of this very distributed storage, which adds an abstraction layer, must also be quickly tossed and turned. And the network must be fast, because the write operations will be block, as I understand it ...

Answer 5

I once solved a similar problem. It was necessary to build a ~100tb file server out of shit and sticks. You can turn up your nose for a long time and argue in the vein: "fe ... you need to build on branded hardware and only ...", but the reality is different. When there is no iron in the city (at all, none) and there is no one in the office to bring iron because because, and a solution is needed - your task is to take and make something out of nothing. On the basis of nas4free on flash drives and a bunch of ordinary sata-screws (only you need a lot of RAM) they raised file storages from a zfs-based backup. This case was given via iscsi to a screw server, which itself lived inside esxi on a gaming mother with i7. There, these storages were collected in one array, which was already given to users. As a result, continuous storage is spread over several file arrays. ZFS gives good speed and fault tolerance. But the Windows server needs network intel (for example, Intel® I350 or Intel ET). Any gigabit dlink / tplink, etc. die by about 200-300Mbps, and esxi does not recognize them either. Network go to psie. Those. interfaces on mother should be more than one. On two ports, data from the NAS is poured, and on two it is poured onto users. Accordingly, the switch is also needed, not the cheapest.