Options for redundant storage cluster?

[Adam Vande More]

On Thu, Feb 18, 2010 at 7:00 PM, Matthew Law <matt at webcontracts.co.uk>wrote:

> Hi,
>
> hopefully I'm not too far out posting this question here.  It takes in a
> lot of areas so I was unsure where to post it.  If it belongs on another
> ML please advise and I will re-post it there.
>
> I am researching options for a two node failover storage cluster. This is
> primarily to provide shared storage (either iSCSI or NFS) for XenServer
> VMs.  I am looking to get the best bang for the buck and wondering if
> FreeBSD might be a good choice?
>
> Hardware-wise we have available two identical supermicro chassis each with
> 16 x SAS bays and a choice of AMD or latest Xeon 5500 CPUs, together with
> as many gigabit cards as we need but the budget won't stretch to faster
> networking.  It would be nice to take advantage of ZFS and use two or
> three 8-port SAS HBAs in each server rather than expensive hardware RAID
> cards.
>
> We don't need to store more than around 2TB but we would like to
> comfortably service around a 75 - 100 VM instances (the VMs on average,
> are not too I/O heavy).  Thin provisioning and snapshots would be nice,
> too.
>
> My initial thoughts were that we might be able to use ZFS, cheap LSI 8
> port SAS HBAs together with a dozen or so SATA II drives and a couple of
> Intel X25E SSDs to help things along.  It would be great if these boxen
> could network boot, so we can use all the drive bays for storage.  I have
> no idea what options exist for clustering NFS/redundancy.
>
> I would be very grateful for any advice - especially from anyone who has
> experience in the same scenario.
>
> Thanks in advance,
>
> Matt.
>

I'd say right now ggated/ggatec + heartbeat is sort of roughly equivalent
of  DRBD and heartbeat.  I think many of us are waiting for HAST though.

http://lists.freebsd.org/pipermail/freebsd-announce/2009-October/001279.html


-- 
Adam Vande More