I think the situation being described here is that the ‘6’ are physical cluster
nodes, not VMs. Vmotion ensures VM integrity while they’re being moved around.
The setup I work with is all based on VMs in an ESX cluster as well, and it’s
great, but I think our plan is to eventually have another cluster hosted in a
separate building, with some kind of regular replication (rsync + slony-I or
similar), so that we not only take advantage of the robustness, scalability and
cool features of an ESX cluster but can feel safe knowing that we’re not
relying solely on that cluster, its SAN, etc., or even that one datacentre.
This stuff is more focusing on ensuring the machine/OS is available, rather
than clustering or load balancing between Dspace nodes. I’m not sure where to
start with the latter, and I don’t recall seeing it discussed.
</2c>
(I also have some ideas related to Amazon’s EC2 service, and having some
offline backup images stored there that can be brought up in the event of a
local failure, but I haven’t tested or investigated it too thoroughly yet)
Cheers,
Kim.
From: John Preston [mailto:byhisde...@gmail.com]
Sent: Tuesday, 21 April 2009 12:35 p.m.
To: Van Ly; dspace-tech@lists.sourceforge.net
Subject: Re: [Dspace-tech] High Availability DSpace
In theory, from reading the documentation in my limited experience, it
allows the clustering of many real hardware elements to support a Virtual
Datacenter/Virtual Infrastructure.
In practice, 1 of 6 in a VMware cluster had a memory problem; we
vmotioned the DSpace to another in the cluster, there was no break in service,
the real host went off-line for servicing. The same was done for upgrading
bios. The vmotion can be set to trigger conditionally. It gives you one more
layer for high availability.
In this case, is it setup so that the 6 vmware virtual machines have separate
DSpace instances, and something like linuxha routes requets to the seperate
instances. If so how do you keep each of the DSpace instances in sync.
John
------------------------------------------------------------------------------
Stay on top of everything new and different, both inside and
around Java (TM) technology - register by April 22, and save
$200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco.
300 plus technical and hands-on sessions. Register today.
Use priority code J9JMT32. http://p.sf.net/sfu/p
_______________________________________________
DSpace-tech mailing list
DSpace-tech@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dspace-tech