On 10/22/12 3:34 AM, Guillaume Belrose wrote:
Hi all,
I've been investigating Pacemaker/Corosync for providing high availability for
a wide range of
applications. I found this combination to be very useful. Some of my
applications require a
fail-over cluster while others require load-balanced cluster.
I am wondering what are the best practices when managing the clusters for those
applications.
Currently, each application runs in a separate/dedicated cluster. I essentially
have different
corosync configurations, one per cluster.
I am wondering if it is not better to setup 1 large Pacemaker cluster which is
partitioned in such a
way that certain resources are dedicated to a certain application (using node
attribute expressions
like in
http://www.clusterlabs.org/doc/en-US/Pacemaker/1.0/html/Pacemaker_Explained/ch-rules.html#s-expression-attribute).
There would only be 1 Corosync configuration, and the cluster is partitioned
using some naming
convention. It seems to me that this would simplify management.
I wonder what people think about this approach.
Many thanks in advance.
Guillaume.
Complexity is the enemy of reliability. Putting all your eggs in a single corosync cluster basket
would lead to lower reliability. These things are always a balancing act between cost (in terms of
hardware resources and sysadmin resources) and reliability. I
--
Ron Kerry rke...@sgi.com
Global Product Support - SGI Federal
_______________________________________________
Pacemaker mailing list: Pacemaker@oss.clusterlabs.org
http://oss.clusterlabs.org/mailman/listinfo/pacemaker
Project Home: http://www.clusterlabs.org
Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
Bugs: http://bugs.clusterlabs.org
