On Mon, 2007-01-08 at 14:33 -0200, Otavio Salvador wrote:
> Tim Post <tim.post@xxxxxxxxxxxxxxx> writes:
>
> >> Consider that, where the block device is primary, the VM is active on
> >> the host, and where the device blocks is secondary, the VM is inactive
> >> on this host.
> >
> > You may find this is easier utilizing network attached (centralized)
> > storage in lieu of block level mirroring, however your explanation is
> > somewhat confusing.
>
> Like use a Storage you even a Cluster FS like GFS or OFS2?
I'm partial to ocfs2, but sure you could use a cluster file system.
There is no need to do this if you're only doing fail over as no two
guests will access the same FS at once.
That being said, jfs is preferable to ext3 in some cases. With a cluster
FS you'll have to deal with the backup node complaining about a cluster
member vanishing (the primary), since it only comes into action should a
primary fail according to your diagram. Just take care to squelch that
with gfs / ocfs2 respectively.
Again, I don't think you need one.. but could use one if you like.
>
> >> We wish information from the list if this solution have been done by
> >> someone,in positive case, if there is some documentation that
> >> contemplate this implementation ,and in negative case, how to do an
> >> implementation of a HA integrated with Xen.
> >
> > Your solution should work exactly as you described it, if I'm
> > understanding it correctly. My only recommendation is to ditch block
> > level mirroring and go with centralized storage.
>
> Yes. But if you're not going to use a clusterfs you'll end using a
> Storage solution that is, indeed, the better option but also more
> expensive. Am I wrong Tim?
>
That really depends on what you consider to be ' expensive '. This
should be determined by how important whatever it is your making highly
available is to you.
If you trust it to 3 - 5 SATA drives and a 3ware / Areca controller than
no, not too terribly expensive. Depends on your I/O needs also.. if
rather light, a lower class Opteron could suffice as a NAS (or its Intel
counterpart, I much prefer AMD but don't want to seem biased).
If you'd rather trust it to a SCSI raid , or FC storage, then yes, its
way more expensive.
Consider time to setup, administrative costs and being the only one in
your office who can fix drbd .. as opposed to $1500 - $2000 more for a
nas. I guess you'd have to weigh it all.
>From an administrative standpoint its easier to manage, offers slightly
better availability depending on the quality of the RAID, and can
increase performance over using drbd + cheap drives and on board SATA
controllers.
AoE would do nicely , I think for your setup and is just a modprobe
away.
Honestly, this is a call you'd need to make .. but weigh more than just
out of pocket costs. ;)
Best,
--Tim
_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users
|
Home