• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Raid 6 large arrays

M

Metraon

Limp Gawd
Joined
Feb 23, 2011
Messages
307
I just landed this quick repair service for a company and they just bought a server with about 160 TB of storage.

They plan to install windows server and do some SMB shares for a windows domain, and the windows part is a requirement for them. But anyway I didnt want to be out of line and I did my tasks, but inside of me I was like 160 TB on raid 6 mean a lot of recovery time in case of failure.

Is it possible just to buy a second server, install windows server on it and install something like nexenta on the first one ?

I think (but I never saw) its possible to do CIFS with ZFS. Will the extra investment worth it ? Will there be performance hits ? They have huge file size, many of the files are 100gb+, they are not that concerned about performance but still they dont want it to be slow poke.

Thanks !
 
I assume there are many smaller arrays that make up the 160 TB. I mean they do not have a 40 x 4TB single raid6 array. That would be insane.
 
I would care about the following on large arrays:

There are silent data errors with a statistical rate. With large arrays this is a real problem to detect and repair these errors. You need data checksums to find the problems with a filesystem that can fix these errors online like ZFS with scrubbing.

With NTFS if you think about Windows, you have no chance to find these errors. In case you need to do a filechck, you cannot repair these arrors and despite that, your array must be offline for a long time for filechecks. You overcome this, you need a CopyOnWrite filesystem (Mainly ZFS but in future btrfs or ReFS as well)

Windows with ReFS adress this problem but its not really ready yet at least not with large Raid-6 arrays or perfomance wise.

What I would prefer is a ZFS box with multiple software Raid-Z2 (3 or 4 x Raid-Z2 vdevs) to increase performance and reduce rebuild time with a hotspare disk with as much RAM as possible as readcache. With multiple users even > 100 GB RAM makes sense on such large files.

When you use a Solaris based OS (Oracle Solaris 11 or Nexenta as a commercial solution or OmniOS as a free option) you can use the Solaris CIFS server that offers you a quite perfect CIFS server that acts from outside view as a quite native Windows server with AD integration, Windows ACL support and Windows SID (your permissions stay intact if you move such a pool) together with ZFS snaps as Windows previous version.

For my own, I use multiple OmniOS as CIFS storage server in an AD environment with 40 TB and about 600 users (University use) and I am happy with this solution that replaced our former Windows filers. Its very stable with nearly no downtime. Only maintenance is a SMB service restart in case the AD server was rebooted.
 
You must log in or register to reply here.
Back
Top