On 6 May 2009, at 13:44, Heiko W. Rupp wrote:
Adrian Cole schrieb:
> This is interesting. What is important, afterall? In the case of
> a grid, it is more like a quorum that allows operations to continue
> without data loss. I'm not sure if individual instances matter as
> complete sets of EC2s could go up or down and there still be no
> effect on cluster as a whole.
This is a good question, that I was thinking about too
> Would it not be the cache instances, or jgroups configuration that
> are the most important managed resource in this case?
Help me with the naming conventions here.
1 Infinispan istance can have n caches, that each has 1 jgroups
channel?
No, they share the same channel.
In the cloud you have x IS instances with n caches.
Yes. Typically 1 infinispan instance per JVM, but not strictly so.
You would have 1 infinispan instance per Channel though.
So in this case, we could have the cache as top level element
below the "IS subystem". And the number of IS instances as
just a numeric metric.
In a later version of Jopr, we will have the possibility of
"related
resources".
In this case we don't need the strict parent-child relation that we
have now.
I think the parent-child relationship works, IMO. Let's not speak of
Infinispan instances since this is a concept that doesn't strictly
exist. A better way to look at it is this:
* You create a CacheManager. This starts a channel. So there is a 1
to 1 relationship here.
* You could start > 1 CacheManager per VM, but I don't see this being
common.
* You use the CM to start > 1 caches. These caches share the same
Channel started by the CacheManager.
* When a channel is created, it has a ClusterName.
So you could organize things by cluster name, and within that, each
cluster has multiple cache managers (named either by IP Address or by
some form of 'node-naming' we can work out), and within that, a set of
named caches.
Cheers
--
Manik Surtani
manik(a)jboss.org
Lead, Infinispan
Lead, JBoss Cache
http://www.infinispan.org
http://www.jbosscache.org