Hi Everyone,

Take a look at this wiki which has details on SC support for LDoms:

http://wikis.sun.com/display/SunCluster/Sun+Cluster+3.2+2-08+Release+Notes#SunCluster3.22-08ReleaseNotes-optguestdomain

The main thing to keep in mind is that you fail-over your applications between 
guest domains with SC. In that regard it's no different than if you have two 
stand-alone servers that are clustered, you fail-over the applications. 

Keeping the LDM configuration in-sync between multiple servers is not possible 
with SC or even VCS. I would see that as a key requirement in doing a full 
blown fail-over of a guest domain.

VCS simply allows you to start/stop a guest domain that's already configured 
between the hosts. You can install VCS in "stand-alone" mode in your guest 
domains and have each guest domain be a "remote" resource in a cluster. So you 
could move a guest domain between hosts, but all the infrastructure (disks, 
networking, etc.) would have to be available across all hosts. It won't do any 
sanity checks for things like mac addresses, ldm configuration, etc. .. so it 
still requires one to be careful.

Now on the flip side, in the future we'll have the ability to move a guest 
domain from one physical host to another without any clustering software. This 
will be known as "Live Migration". It has been mentioned a few times in our 
forum here. But basically, it'll give us a Vmotion type of functionality. That 
should probably be leveraged for moving a whole guest domain between servers. 

It is important to understand the difference between HA clustering and the 
mobility of virtual machines.. they are different things:) Much in the way that 
VMWare Vmotion is not HA clustering.. it's a method of making a virtual machine 
mobile:) It has no ability to monitor your applications and services within 
your virtual machine. This is where HA clustering for applications and services 
should come into play.

 *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
Octave J. Orgeron
Solaris Systems Engineer
http://www.opensolaris.org/os/community/sysadmin/
http://unixconsole.blogspot.com
unixconsole at yahoo.com
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*



----- Original Message ----
From: Ashutosh Tripathi <[email protected]>
To: Paul Roberts <P.Roberts at Sun.COM>
Cc: ldoms-discuss at opensolaris.org
Sent: Monday, September 22, 2008 6:03:54 PM
Subject: Re: [ldoms-discuss] LDOM "Clustering"

Hi Paul,

    Currently, to make a LDoms guest domain HA, you would
run SC inside that guest domain, essentially treating it
as a SC "node". If that "node" fails, all services within it
would be failed over to other "node"s in the cluster and
achieve HA that way.

    What you are suggesting is closer to the "Failover Guest
domain" model (similar to "Failover Zones" model you seem to be
alluding to). That is currently not possible, but we are
looking very actively into it.

Check out my blog entry:
http://blogs.sun.com/SC/entry/ldoms_guest_domains_supported_as

and comments on that entry. It seems several people are interested
in it. Given that you are internal to SUN, can you contact me
offline with your specific customer needs? That would help us
prepare a business case and help move this along?

Thanks,
-ashu


Paul Roberts wrote:
> I've been searching for a while now regarding the best way to have HA 
> LDOMs.  Currently, we are able to implement this solution:
> 
> - 2 Physical Hosts - Node A and Node B
> - Each running Sun Cluster 3.2
> - Using HAStoragePlus agent with ZFS + Zones (SUNWsczone)
> - We're able to fail a zone/container over from Node A to Node B easily 
> and it works great and very easy to maintain
> 
> Is there any way today to fail an LDOM over in the same manner? 
>  Meaning, can I have ldm installed on both nodes and when one node (Node 
> A) goes down or I decide to failover the LDOM automatically comes up and 
> mounts the appropriate storage on the other node (Node B)?  I apologize 
> if I am not being clear enough - feel free to ping me to ask more questions.
> 
> Thanks for your help and guidance in advance,
> 
> Paul Roberts
> -- 
>       ______
>      /_____/\
>     /_____\\ \                Paul Roberts
>    /_____\ \\ /               Sun Microsystems, Inc.
>   /_____/ \/ / /              Sun Serviceability and Readiness Ops
>  /_____/ /   \//\
>  \_____\//\   / /             email:  p.roberts at sun.com 
> <mailto:p.roberts at sun.com>
>   \_____/ / /\ /              direct: 877-477-7642
>    \_____/ \\ \              
>     \_____\ \\                
>      \_____\/                http://www.sun.com <http://www.sun.com/>
> 
> 
> 
> 
> 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> ldoms-discuss mailing list
> ldoms-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/ldoms-discuss
_______________________________________________
ldoms-discuss mailing list
ldoms-discuss at opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/ldoms-discuss



      

Reply via email to