Best options for storing and sharing data over mutiple servers, to be accessed by multiple users

In the environment I currently work in there are several servers dedicated to various departments. However we recently got our hands on several identical servers potentially, around 10+. I'm hoping someone can point me in the right direction on a good option to utilize all of these servers effectively.

Essentially we store large files on the server to be downloaded and less frequently uploaded to the servers.

My goal is to create a good option for the data to be accessible in one (or a few) central location(s) that I can point scripts to and the users can access easily.

What I would ideally like to do, is link up all of these servers as one, or a couple Virtual Machines, or one Networked File Share, that has a good amount of bandwidth and redundancy to keep the files secure and allow the data to be copied down as quickly as possible.

Some things I have been looking into to accomplish these tasks:

-DFS

-Failover Clustering

-Starwind Virtual SAN (The free version)

As a note these servers are quite old, from around 9 years ago.

Any advice on configuring this would be greatly appreciated, I'm a bit out of my depth

Solution 1:

DFS-N is OK (if applied properly, of course!), while DFS-R should be avoided like a plague. It has many issues and inability to replicate open files is a very serious one. It's funny MSFT had a "fix" for this issue at some point, but it never made its way out to the production builds unfortunately.

Windows Server Failover Cluster (WSFC) is a mechanism which is going to require some shared storage (within your scenario), so it isn't a thing on it's own.

StarWind (even free version) will allow you to create a Cluster Shared Volume (CSV) and you can layer WSFC on top of it just fine doing what you want at the end of the day.

P.S. There are other storage options of course, incl. Windows Server built-in S2D, but it requires a Datacenter edition all-around (expensive!) is way less time-tested & reliable.

Solution 2:

DFS replication is a nightmare. For 9 years old servers, I would go with Starwinds VSAN to do the replication between them. That was the reason why I have used it on one of the deployments. S2D definitely won’t fit for this hardware age. You can then combine the nodes into the failover cluster as mentioned and run a file server/VMs on top of a CSV. If you would like to use DFS, I would leave replication to Starwinds and configure DFS namespace on top.