Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

We are looking at upgrading our storage capacity with an external RAID subsystem that has redundant (2) fibre controllers, each controller has 4 x 8 Gbps fibre ports. I would like to make access to this storage system occur via HA Linux.

Ideally I would connect 2 fibre ports from each controller into each Linux server, and then export either NFS or iSCSI via a 10 Gbe interface.

I have seen plenty of references to DRBD, however all of those references tend to use block storage that is solely attached to each machine, rather than having a shared block storage device, so I am unsure if DRBD could (or should) be used in this case.


share|improve this question
Sounds like maybe you should have purchased redundant NAS units with 10GbE interfaces rather than FC units that are using *nix boxes as an intermediary that turns them into NAS storage. – MDMarra Nov 12 '12 at 20:28
redundant NAS units aren't generally able to run in an HA mode, yes, they can backup to each other but if virtual servers are connected to the NAS then when the NAS goes down any servers connected to that NAS are screwed. – nmadura Feb 11 '13 at 20:50
AFAIK, NetApp, EMC, HP, and IBM all offer this functionality is at least some of their product line. I'm sure there are other too, those are just off the top of my head. – MDMarra Feb 11 '13 at 20:56
I wouldn't really call anything NetApp offers a NAS unit, all of the NetApp servers (V series and FAS series) are both NAS and SAN units, and require additional disk shelves (read SAN connectivity) to offer HA type options. I imagine this is also true of EMC, HP, IBM solutions. A fully redundant system from these places is >> $50K and will still use Fibre or SAS interconnects to the shelf storage. – nmadura Feb 11 '13 at 21:15
They all offer CIFS/NFS connectivity to the clustered storage. What exactly is your definition of NAS? The most common definition is that it offers file level access over a network where a SAN offers block storage. If you have unusual requirements you should state them – MDMarra Feb 11 '13 at 21:24

If you just want to present file storage to your network, get NAS units, not SAN.

share|improve this answer
NAS units don't offer the Failover that I am looking for. The block storage is generally inside of them so if you need to take one offline (for upgrade or other reasons) it means bringing the entire filesystem offline for some amount of time. – nmadura Feb 11 '13 at 20:45

You can do this with a combination of a heartbeat solution like keepalived or ucarp... and a virtual IP setup for client access to the two Linux servers. Some form of fencing to use the nodes in an active/standby capacity.

Can you elaborate on your storage server setup and what tools you have available to you?

A more polished option is to use a ZFS filer as a NAS head to present the backend block storage to clients via a variety of protocols (iSCSI, Fibre, NFS, CIFS). NexentaStor is great for this. This is a common use-case for repurposing legacy SAN storage. Plus, you'd have the option of using Nexenta's cluster/HA solution to handle high-availability.

share|improve this answer
I can't elaborate much more at this point. However, this looks like a solution that might work. We are definitely in a situation where fiscal resources are minimal for this kind of stuff. Overtime we have acquired various (non-homogenious) Fibre RAID storage units. Thanks! – nmadura Feb 11 '13 at 20:55

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.