There are 2^28 possible IP multicast addresses, but only 2^23 possible Ethernet (MAC) multicast addresses (http://technet.microsoft.com/en-us/library/cc957928.aspx), which is why e.g. gets mapped to the same MAC address as

This can be inefficient as hosts interested in only the former address will end up receiving packets for the latter as well, i.e. filtering is not done by the switch/router, but in the host itself.

Now IGMP works using IP addresses, and IGMP snooping switches descend to layer 3 anyway. Doesn't this mean that IGMP snooping switches don't suffer from this MAC multicast address ambiguity problem and can route in a way that hosts only interested in don't get packets for


The internet draft IGMPv3 and IGMP Snooping switches believes the following to be most common in IGMP snooping switches:

Group membership is based on multicast MAC addresses only

In other words: The IP-to-MAC multicast address ambiguity still applies to most common IGMP snooping switches.

This makes sense: Routing of normal multicast traffic, i.e. all those multicast packets that do not have the 'Router Alert' IP flag set (which excludes IGMP reports, that do need to be snooped upon and inspected at layer 3) can be done on layer 2, which is more efficient.

However, these CISCO docs indicate configurability:

You can configure the IGMP snooping lookup method for each VLAN. Layer 3 IGMP snooping lookup uses destination IP addresses in the Layer 2 multicast table (this is the default). Layer 2 IGMP snooping lookup uses destination MAC addresses in the Layer 2 multicast table.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.