Priscilla, Presumption is the mother of all evil. I presumed that the switch wouldn't care about MAC addresses on a trunk port. I guess the isl or dot1q encapsulated frame is decapsulated coming in and re-encapsulated going back out (if in fact it goes back out), so there is no reason why the address wouldn't be cached.
I have read in a few different texts that the CGMP message goes out to a well known MAC address. I guess it would have to be forwarded by downstream switches out at least all trunk ports if not all ports for it to work. Thanks much for the clarification(s). I have a bit to learn about the modern switched LAN environment. Mostly a WAN guy trying to branch out. Regards, Scott Priscilla Oppenheimer wrote: > > At 07:27 PM 2/6/02, s vermill wrote: > >Hello all, > > > >I am studying multicast technologies for acedemic reasons and > also so that I > >can hopefully resolve a real-world issue. I have been reading > about CGMP in > >various Cisco Press books, CCO - all the usual suspects. > Seems that all of > >the examples show a switch directly connected to a router. Is > it just a > >foregone conclusion that a Catalyst switch with an on-board > route processor > >will be the architecture? > > No. It could be a "real" router that is configured for CGMP. > > >Because I was thinking of the typical switch > >block. You have your distribution layer switch (hopefully L3 > capable) and > >also a bunch of downstream access layer switches. If a switch > were stacked > >below an upstream access layer switch instead of being > directly connected to > >distribution, I guess CGMP would break? > > I don't think it would break, although there are definitely > many cases > where it does break. Is this sort of what you have in mind? > > Router (or distribution switch with router blade) > | > | > Access switch with CGMP > | > | > Access switch with CGMP > | > | > multicast recipients > > > A host on the lowest switch in the > >stack might send an ICMP join. When the router sends the CGMP > message to > >the all-switch multicast, the higher layer access switch > wouldn't have the > >MAC address of that host stored. > > That's where I would disagree with you. Switches need to store > all MAC > addresses that they can get to, not just the ones that are > directly > connected. I think the higher-layer switch would have the MAC > address of > the multicast recipient that sent the IGMP join (at least if > you were using > my simple topology.) > > My concern is more with the CGMP message to all switches. Do > switches > forward these? They must. > > Well, it's time to get off the computer for tonight! :-) Talk > to you tomorrow, > > Priscilla > > > So it wouldn't forward the multicast frame > >on any of its ports? Or is it sent along all trunk ports by > default? > > > >I sense that I have lost sight of the big picture. > > > >Many thanks, > > > >Scott > ________________________ > > Priscilla Oppenheimer > http://www.priscilla.com > > Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=34725&t=34704 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
