On Thu, 2 Aug 2012 14:56:38 -0700
"Yount, William D" <[email protected]> wrote:
> My expectation with an active/active cluster is that the resources
> are running on both servers at the same time and they are accessible
> from either server simultaneously. If either server goes down, the
> collocated IP address should mean that all the traffic automatically
> flows to the one server that is still up. 
> 
> I am quite confused as to why there is a "process" of switching from
> one node to the other. I am starting to wonder if active/active truly
> means active/active.

Active/Active does indeed mean that the same resources are running on
all participating nodes(*). And you can access the resources on all
nodes. But when you use one "cluster ip", you can have this either on
one of the nodes, then failover takes a short amount of time till its
moved to the other node. Or you have several hosts answer the
cluster-ips connection on a distribution-scheme. But then when one node
fails, there is also a short amount of time till the other nodes
recalculate the answering-scheme. So for some users or ip-connections
there will be a short hang or even a disconnect/reconnect. Which isn't
a problem with small http-requests but your ssh-session will fail...

And the complexity of an active/active cluster are often not worth the
benefits. Most setups are fine with a short outage of up to a minute.
Except windows accessing samba:-(

Have fun,

Arnold

Attachment: signature.asc
Description: PGP signature

_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to