|
On Thu, 2005-12-15 at 12:16 -0800, Mark wrote:
> That sounds exactly like what I mean. Haven't heard of NetApp yet,
> however. Do you know if the same could be done with DRBD or GFS
> (Redhat/Fedora), or do those work differently?
Significantly differently at the layer where the filesystem lives, yes.
I looked into the option of using GFS recently for a specific job and
decided that the convenience of ordering a pair of clustered
"appliances" - whether NetApp, EMC, whoever - with the built-in failover
and fault tolerance won out against building something from the ground
up with GFS, even if the costs were higher in terms of capital layout.
For me to do something with GFS would have required far more dev time
(and therefore extra, "hidden" cost). Plus I have little SAN experience,
and no SAN infrastructure for that job.
Have a look at http://www.redhat.com/docs/manuals/csgfs/browse/rh-cs-en/
to get your head round the concepts.
To appraise the "other side" - vendor-supplied stuff - see:
http://www.netapp.com/
http://www.emc.com/
for starters. And take your chequebook - they most definitely are _not_
cheap!
Graeme
|
