I very badly need to make sure that a server is always present, that VM is
always available. Even if it means dropped calls when failing over to another
server, that would be acceptable.
I've asked about fail over in the past and wondering if anyone has worked on
such a beast to date?
HA to
-users@list.sipfoundry.org
Sent: Sun Jul 18 16:34:43 2010
Subject: [sipx-users] Centralized database and VM
I very badly need to make sure that a server is always present, that VM is
always available. Even if it means dropped calls when failing over to
another server, that would be acceptable
...@grounded.net; sipx-users@list.sipfoundry.org
sipx-users@list.sipfoundry.org
Sent: Mon Jul 19 06:24:41 2010
Subject: Re: [sipx-users] Centralized database and VM
I don't think centralized voicemail has been addressed yet.
In theory in a HA environment it could cause a lot more WAN traffic so its
probably
m...@grounded.net m...@grounded.net 07/19/10 7:25 AM
HA to me always means having fully separate servers without one main
controller.
For example, web servers should have fail over front end load balancers to
multiple back end servers.
In the case of sipx, even some fail over would be good.
On Jul 18, 2010, at 3:34 PM, m...@grounded.net m...@grounded.net
wrote:
I very badly need to make sure that a server is always present, that
VM is always available. Even if it means dropped calls when failing
over to another server, that would be acceptable.
I had considered going with
I have been wondering about a combination of what you guys are doing.
Rather than keeping two servers in sync, my thought is running two base
servers, where each has it's sipx related directories on a very fast network
share. The database server would be hosted on another server and a client
From: sipx-users-boun...@list.sipfoundry.org
[sipx-users-boun...@list.sipfoundry.org] On Behalf Of Tony Graziano
[tgrazi...@myitdepartment.net]
I don't think centralized voicemail has been addressed yet.
In theory in a HA environment it could cause a
On Mon, Jul 19, 2010 at 1:13 PM, WORLEY, Dale R (Dale)
dwor...@avaya.com wrote:
VM is a particularly difficult situation. It is an operational component
(as opposed to a configuragion component), so it has to operate in more or
less real time.
Are there situations particularly in branch
I think there's more to it than that.
Right now, even in HA, wherever the user is created is where their VM box
is. So there are some dependencies on the sole fact that if the server has
VM, then that is where the user needs to check it. So it is more than an
asynchronous replication, but also
There's only one way to find out what might work, I'm going to build a couple
of systems and see what I can come up with.
Sometimes, solutions are found by not over complicating the thinking first,
trying it, then working through the problems.
What would happen if you had two sipx systems,
What about something like using Exchange Unified Messaging, and the
clustering you can do with that for voicemail? It's not open source,
but if you were a heavily Microsoft shop already, it might be
something that was withing reach.
Call control could stay with Sipx, and the voicemail records
From: sipx-users-boun...@list.sipfoundry.org
[sipx-users-boun...@list.sipfoundry.org] On Behalf Of Douglas Hubler
[dhub...@ezuce.com]
There would always be a window where vm could get lost though unless
someone implements real-time replication filesystem
I think in his instance he is hosting service for lots of folks, so Exchange
becomes unwieldly and expensive.
On Mon, Jul 19, 2010 at 4:35 PM, Kurt Siegfried kurt.siegfr...@gmail.comwrote:
What about something like using Exchange Unified Messaging, and the
clustering you can do with that for
Given that status information is stored in the VM database, I would
expect that a filesystem replication strategy would work 98% of the time,
and the other 2% would destroy information. I wouldn't trust anything less
than a real distributed DBMS to work robustly.
Thing is, when the standby
14 matches
Mail list logo
