IGMP Snooping

[subbarao.s]

I think IGMP Snooping is a bit in-effecient,Correct me if I am wrong.

I have a cat3550-24EMI I wanted to take adavantage of IGMP Snooping. I have a mrouter attached to f0/1 running PIM Dense mode and i have client on f/02.

When I join a group 226.1.1.1 from client I can see Multicast MAC entry as below

ABC#sh mac-address-table multicast

Vlan Mac Address Type Ports

---- ----------- ---- -----

1 0100.5e00.0128 IGMP Fa0/1

1 0100.5e01.0101 IGMP Fa0/1, Fa0/2

After that I added another group 227.1.1.1 from Fa0/3, then I can see same MAC is mapping to all 3 ports. like below

ABC#sh mac-address-table multicast

Vlan Mac Address Type Ports

---- ----------- ---- -----

1 0100.5e00.0128 IGMP Fa0/1

1 0100.5e01.0101 IGMP Fa0/1, Fa0/2, Fa0/3

Is this not ineffecient? traffic designated for one group floods to 2 ports with dissimailar group clients.

I am just curuios whether this is RFC standard to take only final three octets while making this multicast MAC or only Ciscos implementation is so.

Thanks.

[milan.kulik]

Hi,

see RFC1112 (http://www.cis.ohio-state.edu/cgi-bin/rfc/rfc1112.html) , paragraph 6.4.:

"An IP host group address is mapped to an Ethernet multicast address by placing the low-order 23-bits of the IP address into the low-order 23 bits of the Ethernet multicast address 01-00-5E-00-00-00 (hex). Because there are 28 significant bits in an IP host group address, more than one host group address may map to the same Ethernet multicast address. "

I've also found a funny story about the reason (http://www.certtutor.net/Forums/messageview.cfm?catid=75&threadid=32191):

"Specifically, I think the story goes that the guy who did much of the original work on multicasting, Dr. Steve Deering, wanted to buy some MAC-OUI-address-blocks for his multicast addresses. However, Dr. Jon Postel (along with Dr. Vint Cerf, the grandfathers of the Internet), who was Deering's manager, had very little money (remember, these were the years when the Internet was not commercialized), and so could only give Deering enough money to buy one OUI block. That block is the one that starts with 01:00:5e, and that is why multicast addresses start that way. Furthermore, Postel had to use half that block for something else (again, it was a money thing), and that is why only half that OUI block is available for multicast."

Regards,

Milan

[subbarao.s]

Milan,

Thanks alot. That answers my question...Interesting stroy though (-:

Regards,

Subba Rao.S

[sbhadrav@cisco.com]

Multicast-capable router ports are added to the forwarding table for every Layer 2 multicast entry. The switch learns of such ports through one of these methods:

•Snooping on IGMP queries, Protocol Independent Multicast (PIM) packets, and Distance Vector Multicast Routing Protocol (DVMRP) packets

•Listening to Cisco Group Management Protocol (CGMP) packets from other routers

•Statically connecting to a multicast router port with the ip igmp snooping mrouter global configuration command

You can configure the switch either to snoop on IGMP queries and PIM/DVMRP packets or to listen to CGMP self-join or proxy-join packets. By default, the switch snoops on PIM/DVMRP packets on all VLANs. To learn of multicast router ports through only CGMP packets, use the ip igmp snooping vlan vlan-id mrouter learn cgmp global configuration command. When this command is entered, the router listens to only CGMP self-join and CGMP proxy-join packets and to no other CGMP packets. To learn of multicast router ports through only PIM-DVMRP packets, use the ip igmp snooping vlan vlan-id mrouter learn pim-dvmrp global configuration command.