Raid or Storage spaces, anybody taken the plunge yet?

[Jaan]

how can you do a raid 6 equivalent on one single server

 

The fault tolerance is across the nodes..... I agree, i like the idea on local raid10 on each server clustered with S2D. Maybe MS will implement that someday.

 

some nice pics at the bottom of this page:

 

https://docs.microsoft.com/en-us/windows-server/storage/storage-spaces/storage-spaces-fault-tolerance

[Jaan]

I just spotted this thread..

 

Sorry just noticed your question is with regards to Storage Spaces not Storage Spaces Direct.

 

The only thin i have on that is this article from microsoft about Storage Spaces fault tolerance non Direct.

 

https://technet.microsoft.com/en-us/mt651669(v=ws.11)

[markm75]

Sorry just noticed your question is with regards to Storage Spaces not Storage Spaces Direct.

 

The only thin i have on that is this article from microsoft about Storage Spaces fault tolerance non Direct.

 

https://technet.microsoft.com/en-us/mt651669(v=ws.11)

 

Confused slightly.. so storage spaces direct you dont need other servers or nodes for fault tolerance.. ie: mirroring can be done.. so jbod 10 drives, 5 drives in each mirror i assume?

 

Also confused, some seem to praise storage spaces with 2016+, while others bash it as not ready for prime time or production use.. but if you are doing DPM 2016 and want space savings you need Storage spaces and refs (they go together i think).. has me leaning towards just doing raid10 or raid6 again on the new DPM server box just to be safe, dont gain any space savings with 2016 though.

[ChrisMiles]

In my opinion ReFS isn't fit for many production uses yet, it's far too limited support wise. Even for Hyper-V we had to switch back to NTFS because using ReFS on cluster shared volumes means that they can only operate in the mode where all IO is sent to the controlling host rather than direct access which kills performance. It also severely limits your backup options, breaks DFS replication and data deduplication. I started off trying to use it for as much as possible in a brand new 2016 domain setup and ended up having to switch back to NTFS for everything including hyper-v, file hosts, sql hosts, etc, except for veeam data storage, due to some incompatibility or another. I strongly advise against using it on anything except s2d.

[BKGarry]

Sorry late to the conversation.

 

We actually have a hybrid of the 2 here.

 

So there is a raid for SSDs (mirrored) and a raid 5 with some 4tb disks. These are hardware controlled raids, then in server 2012 r2 we used powershell to declare the SSD RAID as SSD and then Storage spaced them together.

The reason, basically one big RAID SSHD, to which Storage spaces makes sure the most active blocks (and yes it is at block level) are on the SSDs and all that lovely archived, not really used stuff is on the HDDs, but they are still higher performance HDDs not commercial ones.

[krb548]

I have recently been looking at Microsoft storage spaces as a Dell partner has recommend it over the traditional San storage for hyper-v.

We are currently looking at renewing our server environment and replacing all servers with two new hyper-visor hosts and creating a whole new virtual server environment. We have been looking at the traditional method of hosts, fibre switches and san for shared storage.

 

Has anyone taken the plunge yet and used storage spaces instead of the old traditional methods?

To me it is looking more cost effective using the new system but it also looks too good to be true. Is this tech going to be the next VHS & Betamax war and if you choose the wrong one you are stuck for years with failing support?

 

It would be interesting to hear other peoples thoughts

[SHimmer45]

i think "failing support" isnt going to be any issue, the fact that the tech has survived through two versions of Windows server imo they are going to back it to the hilt.