[cisco-voip] QoS in distributed cluster

[Ruben Montes]

Thanks for you response.

Yes, I'm using L3 in the devices: MSFC3 in the 6500, the 4232L3 in a
4006 with CatOs and a C4506 with L3.
Now, my RTT aprox is 12ms, but I would like to configure QoS in all the
network to have all the voice and intra-cluster traffic controlled.

I would use autoqos in the C45006 because this is the only device that
supports that functionality. I would do a similar mapping in the 4000
and the 6500. Do you think it would be enough?

Regards,

Ruben

-----Mensaje original-----
De: Ray Burkholder [mailto:ray at oneunified.net] 
Enviado el: martes, 13 de diciembre de 2005 13:10
Para: Ruben Montes
Asunto: Re: [cisco-voip] QoS in distributed cluster

Hi There,

As an AVVID practitioner, I would highly recommend you review Cisco's
DQOS
course notes and text book (if you havn't already).  They explain the
queueing
structures available in switches and how to configure them in light of
prioritizing prec 3 and prec 5 traffic.  In summary though, with
properly
configured switch QOS, you would be making use of 2 or 4 or more queues
in each
switch type, and they take care of queueing requirements.  I find the
auto-qos
functions in each switch type very handy in figuring out the
peculiarities of
each switch.

In addition, for the WAN or MAN or WLAN that you have, are you using
layer 3
switching in the devices?  Recent Cisco best practices indicate that
alot of
segmentation is a good thing.

One more thing, for distributed clusters, Cisco requires that one way
between
callmanagers must not be more than 20ms, or 40ms round trip.

Many of the Cisco devices have SLA/SAA/RTT capabilities so you can
monitor this
in real time.

In addition, by using some Cricket addons, you can monitor your network
wide QOS
statistics quite readily for delays, jitter, drops, and losses.

Quoting Ruben Montes <Ruben.Montes at eu.didata.com>:

> 	Hello,
> 
> 	 
> 
> 	I have a distributed cluster. The publisher is located in one
> city and the subscriber is in another one. The network is a LAN-to-LAN
> with all L2 interfaces.
> 
> 	In the best case, there are two 100MB hops between the
> CallManagers, and in the worst case (redundant path) two 10MB hops.
I'm
> a little bit worried about that and I would like to configure QoS. We
> have a bunch of devices between the CCMs:
> 
> 	-         C6500 with sup720
> 
> 	-         C4500 with 4232L3
> 
> 	-         C3548
> 
> 	-         C4506
> 
> 	 
> 
> 	My question is if it enough to mark the interesting traffic and
> map it to a high priority egress queue in all the path, or if it is
> necessary to agreggate policiers to limit the traffc.
> 
> 	 
> 
> 	If I had a WAN I would configure CWFQ, but in this case, which
> do you think is the best option?
> 
> 	 
> 
> 	Regards,
> 
> 	 
> 
> 	Ruben
> 
> 
> -- 
> Scanned for viruses and dangerous content at 
> http://www.oneunified.net and is believed to be clean.
> 
> 


-- 
Ray Burkholder
http://www.oneunified.net
ray at oneunified.net
441 505 7293

-------------------------------------------------
Sent from http://www.oneunified.net via IMP: http://horde.org/imp/

-- 
Scanned for viruses and dangerous content at 
http://www.oneunified.net and is believed to be clean.