cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
644
Views
0
Helpful
1
Replies

Multicasting in Campus network issue

Benoit.Cote
Level 1
Level 1

Hi guys,

Let me just give you a quick overview of my issues with Multicasting in my current Environnement.

We are deploying this year Windows 7 through the Windows Deployment Services. I have no issues with Unicast connections to the server, but I seem to have instability when I try to Multicast.

The switches that I use in almost all my labs are Catalyst WS-C2950T 24PS with the c2950-i6k2l2q4-mz.121-22.EA14.bin or c2950-i6k2l2q4-mz.121-22.EA12.bin.

All my distribution switches are Catalyst 3550(fiber) and all have ip multicasting-routing enabled, igmp snooping, ip pim dense mode, etc. In this case, I have a 2970G connected to my Distribution switch(3550) in which all the labs connect to.

Basically, multicasting works for almost any switch that i've got except the 2950's... and yet they are suppose to work from what I have read. I did a test in one of my labs and put a 3560 48-PoE configured the same way as my 2950 and it worked well.. simple configuration on it.. here is an example of the port config on the ports.

description "D3300.1"
switchport access vlan 222

switchport mode access

spanning-tree portfast

I tried putting my uplinks in Trunk mode or access mode.. and still I get issues with the 2950's... What happens is they drop my connections on the deployment server after 3-4-5 minutes or sometimes keeps them at extreme slow speeds... But when I put in a 3560 with the same config... steady connection with a really good speed.. Same thing for the techs in their office.. They deploy some PC's on a 3560G and steady connection... but as soon as I put in the 2950's.. I get issues with multicast.

I know this might not be a great explanation of my configuration but whatever information you need, I'll provide if it can help us fix this issue. We have an upcoming deployment in 2 weeks and I would like it to go smoothly.. and not have to change my 2950's with other switches.(that I don't have hehe)

Could the 2950's choke somehow and decide to cut the multicast? Because I don't lose the connection completely.. the PC's just basically go into Unicast and continue on their way... but as you know..30 PC's in Unicast compared to 30 in Multicast really isn't an option.

To put it in an order for you guys here is what the connection in between one of my labs and the deployment server looks like :

Deployment Server connected on 2970G --->Kore-6509-E ---->Distribution switch D building-3550---> 2970G --->uplink lab switch-2950

So basically from the deployment to the lab, I've got 5 switches total.. hopefully this can help you guys have a little visual of our setup here.

Any help or ideas are welcomed.

Thank you

Benoit

1 Reply 1

lgijssel
Level 9
Level 9

As your problem solely relates to -older- 2950 switches, it might be an issue with the IGMP version. Do you know what versuion of IGMP you are using on the various models?

IGMPv1/IGMPv2 hosts do not understand the IGMPv3 query or IGMPv3 membership query. The router must only use the IGMP version that corresponds to the lowest IGMP client version present. If there are IGMPv3 and IGMPv2 clients, the router uses IGMPv2. If there are IGMPv1, IGMPv2, and IGMPv3 clients, the router uses IGMPv1.

Please check this link:

http://www.cisco.com/en/US/products/hw/switches/ps708/products_tech_note09186a00800b0871.shtml#igmp

regards,

Leo