[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-users] best setup for live failover


  • To: Donny Brooks <dbrooks@xxxxxxxxxxxxxxxx>
  • From: Stephen Ross <stephen.ross1986@xxxxxxxxx>
  • Date: Wed, 11 Aug 2010 01:04:18 +0100
  • Cc: xen-users@xxxxxxxxxxxxxxxxxxx
  • Delivery-date: Thu, 12 Aug 2010 06:31:16 -0700
  • Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type:content-transfer-encoding; b=EwkN8xncJB+rR14Tgv9+DPCDOpaY/iXCXU18AU+51HyB4/gEr4e7PvQvmsxe2FD7/S DCeTdgx5P9H9uZGVO9z3UhQ5Zabh9JYFIW7vejOzx6GrIqMqIfww2L6fwlRR2mIDyo/r nFOUxFZj01ltLIxOA/ZxZGuc7jqUM2FK0FQq4=
  • List-id: Xen user discussion <xen-users.lists.xensource.com>

iSCSI will give you very good performance. Much better than NFS, for
instance. MD3000i's are a bit of a pain though, multipathing these
units using dm-multipath doesn't work very well in my experience. The
latest versions of RHEL do have improved support for MD3000i's though,
very much depends on the Dom0 linux distribution you're using?

Personally, in this sort of setup I've always gone for a cLVM based
approach. Means you don't have to faff about with clustered
filesystems and should be as scalable as you could ever need! Last
role I was in I had 15 MD3000i's and 6 R900's in this configuration
supporting ~300VM's and it was very fast and stable.

As for HA failover, I've used heartbeat for this in the past with good
results but with Xen 4.0 there are various options I'm not experienced
with. RDMA looks great, for instance. Not sure how mature it is yet
though.

On Tue, Aug 10, 2010 at 9:44 PM, Donny Brooks <dbrooks@xxxxxxxxxxxxxxxx> wrote:
>  Well the disk box we are looking at is the Dell MD3000i so iscsi is an 
> option. Does xen handle that better than other protocols? I am kinda new to 
> the whole shared network disk stuff.
>
>
> --
>
> Donny B.
>
> On Tuesday, August 10, 2010 03:04 PM CDT, Bart Coninckx 
> <bart.coninckx@xxxxxxxxxx> wrote:
>
>> On Tuesday 10 August 2010 20:53:07 Donny Brooks wrote:
>> >  Also let me explain my planned roadmap. Instead of buying huge Dell T710
>> >  style servers with cpu, ram, and disks, from this point on I would like to
>> >  just have a NAS or two and then have some small 1U server that have the
>> >  cpu and ram to run the guests on. That is the main reason for the live
>> >  failover.
>> >
>> > > I am about to get a few extra servers and hopefully a NAS for here at
>> > > work. Currently all of our VM's are on one host that has a raid6 with
>> > > approx 12TB usable. The way I would like to set things up is to where the
>> > > disk images (we use lvm partitions for guest disks) sit on the NAS and
>> > > use something like DRBD or similar to allow live failover. We are using
>> > > Xen 4.0 on a Centos 5.5 Dom0 if that matters.
>> > >
>> > > So how would you recommend setting up the system to allow shared storage
>> > > (NAS) with live failover using lv disks? I wouldn't need to failover
>> > > every guest, just the primary ones like mail, web, ldap. Thanks for any
>> > > input.
>> >
>> > _______________________________________________
>> > Xen-users mailing list
>> > Xen-users@xxxxxxxxxxxxxxxxxxx
>> > http://lists.xensource.com/xen-users
>> >
>>
>> For DRBD you need a something to do the DRBD and you would need two NAS 
>> boxes.
>> How about using your NAS as an iSCSI box?
>>
>> _______________________________________________
>> Xen-users mailing list
>> Xen-users@xxxxxxxxxxxxxxxxxxx
>> http://lists.xensource.com/xen-users
>
>
>
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@xxxxxxxxxxxxxxxxxxx
> http://lists.xensource.com/xen-users
>

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.