Questions about Catalyst 3560 and Microsoft NLB

Unanswered Question
Dec 16th, 2009

Hello folks,

I've been setting up a IGMP based multicast Microsoft NLB setup. See my attached drawing for the setup. (Please note that in this setup there is only 1 server - I know that )

I've done it according to the following document:

I have a question regarding the configuration of the two switches. The document above states that in order to do the switching in hardware instead of process switching one must enter the following command:

mac address-table static 0100.5e7f.0b0f vlan 20 interface GigabitEthernet0/12

But as my setup is two switches (with ip routing enabled) and HSRP.

Must I also do the following command in the opposite HSRP switch in order to enable hardware switching in that switch as well? Note that I've pointed the interface to the trunk instead of G0/12.

mac address-table static 0100.5e7f.0b0f vlan 20 interface GigabitEthernet0/24

Thank you in advance!

P.S. The document also states:

Cisco recommends that you do not use multicast MAC addresses that begin with 01 because they are known to have a conflict with the IGMP setup

Can someone elaborate on that and perhaps a fix?

Best regards,

Jesper Petersen

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Giuseppe Larosa Wed, 12/16/2009 - 13:06

Hello Jesper,

here the idea is to build statically the Layer2 replication list or oilist and it works on multicast MAC addresses.

according to ethernet standards multicast frames have the less significant bit of most significant octet set to 1 so any MAC address that starts with

a byte like  xxxxxxx1 is representive of a multicast L2 MAC address.

To be noted that multicast MAC addresses reserved for carrying IPv4 multicast addresses are only a subset of this space.

Multicast MAC addresses for IPv4 start with:


So the Cisco recommendation has to be read as:

to make this trick to work avoid to map the Virtual IP address that is an unicast IPv4 address to a MAC address that is part of dedicated subset for Multicast IPV4 or some IGMP checks will cause the trick to fail.

So: a multicast MAC address out of subset for IPv4 multicast is mapped to the ports where servers are connected/reached.

A static ARP entry that maps the VIP unicast IPv4 address to the multicast MAC address is part of the solution.

To be noted IGMP does not play any role in this trick, rather it can cause problems if you try to use a MAC address like the one you would like to use.

Intermediate switch should have the command for both ports that towards sw1 and that towards server, but you need two servers for doing a meaningful test.

Hope to help



This Discussion