Lars Marowsky-Bree wrote:
On 2006-05-19T06:40:54, Alan Robertson <[EMAIL PROTECTED]> wrote:
Normally, you want only one subcluster - total. We are not doing
clusters of clusters here. Just one single cluster which spans sites.
Also known as a split-site cluster.
(Actually, I believe they are more commonly known as "stretch
clusters".)
Right. That's a better name for it.
I believe it is more appropriate to treat these clusters as
stacked/layered clusters.
You mean a clusters of clusters I think? My first guess is that this
will likely have to wait until release 3.x. One could design a new
separate system for this on top of our current code if one had the
resources. We don't.
This is the way how for example Novell's Business Continuity Cluster
product models this, and also, to my knowledge, Veritas's solution,
Novell's BCS cluster products are a separate product on top of an
existing clustering product. I'm not up for designing a new product at
the moment. AFAIK, we aren't precluding the development of such a
product should resources become available.
whose name I forgot. Stretch clusters are, to the best of my knowledge,
somewhat limitted because they essentially pretend that it is a flat
structure, but it isn't in practice...
What kinds of limitations did you have in mind here?
[Resource level quorums (quora?) would be an even nicer extension, that
would be awesome for stretch clusters but I'm not yet ready for that
(nor ready to ask Andrew for it) either]
Anyway, treating it like this also helps with other forms of
layered/stacked clusters, say virtualization.
In my understanding, this would require a significant rearchitecture of
pretty much all components (CCM, CRM/pengine, mgmtd, heartbeat, etc).
My best guess is that it would be a larger effort than R2. I don't have
the resources to do this at the present time.
Have you had a chance to read the docs I put up on the wiki a week or so
ago? Do you have specific limitations of those proposals that you think
are important?
The implementation currently underway (in early phases) requires a new
quorum tiebreaker modules (and daemon), and perhaps a quorum module.
This is (and a few other variations of it) are described on the web
site. This is a low cost option compared to the probably 5 staff years
to write a separate cluster-of-clusters system, or the probably 7-10
staff years to make it fully integrated and seamless.
--
Alan Robertson <[EMAIL PROTECTED]>
"Openness is the foundation and preservative of friendship... Let me
claim from you at all times your undisguised opinions." - William
Wilberforce
_______________________________________________________
Linux-HA-Dev: [email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/
