On 2009-11-04T08:57:29, lepace <lep...@163.com> wrote:
>
> Hi,all
> I want to configure a HA cluster which having more than 60 nodes,and I
> want to use N-to-N mode,so every node can potentially be used for
> failover.which I want to know is how to configure the value of
> parameters in openais.conf，or the methods to find the right value of
> parametres
I think that if you were aware of a few design constraints, you would
not want to do that.
Consider - a single cluster is a single failure domain. Normally, most
components only have simple redundancy - ie, two of each criticial
component, to eliminate Single Points of Failure.
Now, the likelihood of two components failing at the same time increases
with the number of components (nodes) in the cluster. Would you build a
60 drive RAID-5, which only can compensate a single drive failure? No,
you'd build several smaller RAID5 groups, possibly 12 with 5 drives
each (or larger RAID6 groups, but you get my point).
While we're moving into that direction, the current HA cluster efforts
are not really suitable for large node count HPC environments. I
personally think the sane boundary node count is in the range of 7-16
nodes - we regularly test with 16 nodes.
There's also the point which Steven raised, that the totem protocol
(because of its very much timeout based nature) doesn't scale easily to
large configurations yet.
Regards,
Lars
--
Architect Storage/HA, OPS Engineering, Novell, Inc.
SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg)
"Experience is the name everyone gives to their mistakes." -- Oscar Wilde
_______________________________________________
Pacemaker mailing list
Pacemaker@oss.clusterlabs.org
http://oss.clusterlabs.org/mailman/listinfo/pacemaker
