1

A layer 2 Ethernet frame is sent with a source and destination MAC address. Given a 100Base-TX (copper) trunk between two Cisco switches, I can do a "show interface fa 0/0" on S1 to see the MAC address assigned to the trunking interface, then go to Switch2 and do a "show mac address-table" and find the MAC address of the S1 fa 0/0 interface as a dynamically learned MAC in the table.

Given a similar setup with a 1000Base-X (fiber GBIC) trunk, the MAC address shown in "show interface gi 0/0" on S1 does not show up in the MAC address-table of S2. Everything I can find online indicates that 1000Base-X uses largely the same layer 2 format as copper connections. There's some slight alterations - minimum frame size is slightly larger - but the fundamentals of the frame structure appear to be the same, including transmission with a source and destination L2 address.

Why doesn't the address of gi 0/0 show up in the MAC address table of the connection switch? The only thing which seems to make sense would be that the GBIC has its own MAC address, almost as if its acting as a mini 2-port switch or hub, with the switch-assigned MAC address showing up on the interface connection and a different MAC address assigned to the fiber side. If this is the case, is there any way to see the GBIC MAC address on the switch?

(I've tried to look up the details in IEEE 802.3z but it doesn't seem to be available without an IEEE membership or purchasing the standard. I find the base 802.3 PDFs for download, but not 802.3z.)

user69971
  • 21
  • 1

1 Answers1

2

Your Switch2 mac address-table will only populate with source address entries that it has observed on an ingress Ethernet frame. What's most likely happening is that at some point the Switch1 100Base-TX port's MAC was the source address for an Ethernet frame, while the 1000Basex-X MAC has not been. It could be related to the behavior of the switch in how it sources it's self originated traffic like BPDU frames, etc. When forwarding transient layer 2 traffic, which is to say traffic it is simply carrying along the path, the individual Ethernet addresses of physical interfaces on the switch are not included.

My guess would be that Switch1 sourced an Ethernet frame as the 100Base-TX port MAC at one point, while it has not sourced anything from the 1000Base-X port MAC. If it never sources traffic with the 1000Base-X port MAC, your Switch2 mac address-table will never populate with it.

mcmeel
  • 526
  • 2
  • 6
  • Thanks for the response. We have a large network (over 1200 devices at this location) and I've checked a number of switches - about ten or twelve of each type. In every case where we have a copper trunk, I find the MAC addresses of the trunk ports in the MAC address table of the opposite switch. In every case where we have a fiber trunk, I do not find the MAC address on the opposite switch. The switches are running DTP, STP, CDP, etc., so there is definitely traffic flowing directly between the switches, which I would assume would use the switch interface as the source port. – user69971 Feb 08 '11 at 18:46
  • You could list all the MAC entries in the forwarding table for the peer interface on Switch2 and, through a process of elimination, perhaps identify if an address other than that is listed is in fact learned. I'd imagine the address would use the same first three octets as the MAC listed for the interface. Worst case you can do some clever things with this: http://standards.ieee.org/develop/regauth/oui/oui.txt – mcmeel Feb 08 '11 at 21:27