[cisco-voip] Multicast MoH Delay

[Tony Underwood]

I'm the furthest thing from a Multicast expert, but to my knowledge the static join is joining only the router interface to the multicast stream and it forwards the packets continuously.  But, when the packets get to the L2 switch, it doesn't have an active IGMP join in it's table so it doesn't forward the traffic out any ports.  Then when the phone joins the group it is instantly available at the access layer due to the static join.
So, if anything this tells you that your delay is between the L3 devices and not at the access layer.

Tony Underwood CCIE #7112
Sr. Network Engineer
Cambium Data Inc.
5050 So. 111th St.
Omaha, NE 68137
(402) 556-1388
http://www.cambiumdata.com<http://www.cambiumdata.com/>

From: cisco-voip-bounces at puck.nether.net [mailto:cisco-voip-bounces at puck.nether.net] On Behalf Of Daniel
Sent: Wednesday, June 24, 2009 5:44 PM
To: cisco-voip at puck.nether.net
Subject: Re: [cisco-voip] Multicast MoH Delay

Thanks all for your replies

I would rather use sparse mode to take use of the join messages etc.. so that the phones join and leave the group. Dense mode I think floods the network and then prunes interfaces that are not in use, this occurs every 3 minuts or so. It just seems that sparse mode is more precise, I don't have much experience in this so not sure. Any thoughts / technical reasons on why to go Sparse, Dense or both?

The RP is on a 6500 with sup720s and MSFC3's, the other distribution switch is an older 6500 with SUP2s and MSFC2, so c3voip similair to the issue you had with TAC but the RP is on a different device.

I have added the "ip igmp join-group group-address" command in, this fixes the issue. My question is with this command, the router will accept and forward these packets preventing fast switching, if i use the static-group command the router doesnt accept the packets itself but forwards them thus allowing fast switching, anyone know of benefits either way? I did originally have a look at this command but I assumed its use was to always have the multicast traffic flowing which I didnt think ideal. But it turns out after actually trying this from Tony's point below it works quite well. Between the dsitribution switches the mroute is always setup but not from the distribution switch to the floors, when the phone joins the group the phones vlan interface is added to the mroute and MoH is heard straight away. There is no multicast MoH flooding the floor vlans which is what I was concerned about.








On Thu, Jun 25, 2009 at 5:49 AM, Tony Underwood <tony at cambiumdata.com<mailto:tony at cambiumdata.com>> wrote:

If it's a delay in the route set up then you could try a static igmp join on the far end router.

ip igmp join-group group-address



Tony Underwood CCIE #7112

Sr. Network Engineer

Cambium Data Inc.

5050 So. 111th St.

Omaha, NE 68137

(402) 556-1388

http://www.cambiumdata.com<http://www.cambiumdata.com/>



From: cisco-voip-bounces at puck.nether.net<mailto:cisco-voip-bounces at puck.nether.net> [mailto:cisco-voip-bounces at puck.nether.net<mailto:cisco-voip-bounces at puck.nether.net>] On Behalf Of Daniel
Sent: Wednesday, June 24, 2009 1:46 AM
To: cisco-voip at puck.nether.net<mailto:cisco-voip at puck.nether.net>
Subject: [cisco-voip] Multicast MoH Delay



Hi All,



I've setup multicast routing for music on hold between our data centre network and building floor subnets. The setup is that the data centre is on a different subnet and gear to the floor subnets so multicast routing needs to be used to get MoH packets to the phones.



This consists of the following traffic flow,



MoH Server > Access Switch > Distribution Switch/router(RP) > Distribution Switch/router > Floor Switches > Phones



There are three routing hops (1) from vlan interface to routed interfaces of distribution switch (2) between Distribution and (3) from routed interface of distributionswitch  to phone vlan interface. Only the distribution switches are layer 3.



The first distribution switch is the RP for the group and only that group.The second distribution switch is accepting auto RP only. We are using sparse mode.



Multicast traffic is working, the RP mappings are there, the mroute is there, the problem is that when a call is placed on hold there is a 5 second delay before the music is heard. I assume this to be somewhat because of the join message and the delay of the route or path being setup.



Anyone know of a way to reduce the delay before music is heard? If not I guess its back to the lab.



regards,



Dan



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/cisco-voip/attachments/20090624/c91ab828/attachment.html>