At 11:21 AM 5/31/02, FRANKD wrote: >Hi Priscilla, > >What I did find was that in multi-cast token ring, it uses functional >addressing. Search Cisco for token ring functional addressing. > >Try this one in particular: > >http://www.cisco.com/univercd/cc/td/doc/product/software/ios113ed/113ed_cr/np1_c/1cmulti.htm#xtocid2250942 > >Good Luck, > >Frank
Thanks for the answer Frank. Please send answers to the group, not to my individual address. (I only do multicasting! ;-) That document has to do with IP multicast. I wasn't asking about IP multicast, but the fact that IP multicast uses a functional address on Token Ring is a good hint. Rumor has it that you need to know that for some Cisco tests. ;-) Here's the issue: IEEE says that the first bit transmitted in a data-link-layer header is the Individual/Group (I/G) bit. 1 = group. 0 = individual. Group also means multicast. If all bits in the destination address are 1s, then it's a broadcast, a special case of multicast or group. In canonical Little Endian format (Ethernet), the first bit transmitted is the least significant, low-order, right-most bit. For example, CDP sends to the multicast address 01-00-0C-CC-CC-CC. Notice that the least significant bit in the first byte (01) is a 1. In non-canonical Bit Endian format (Token Ring), the first bit transmitted is the most significant, high-order, left-most bit. I'm glad to see that Wendell Odom removed the discussion of Little Endian versus Big Endian from the latest version of the CCNA book, because the previous edition had it gunked up. Now, back to the real question. I know that IEEE says that 802.5 Token Ring supports multicast, i.e. the setting of the I/G bit. In fact, the IBM TR Reference Manual supports it too. However, Token Ring NIC vendors didn't support it in the past. This caused problems for IP Multicast, DECnet, AppleTalk, and many other upper layers that use multicasting. Some Cisco classes and books teach multicast as if it's only an IP thing. That's wrong. On Token Ring, a lot of upper layers had to rely instead on Token Ring functional addresses. Token Ring says that if the first bit transmitted of the third byte is a zero, then it's a functional address. Functional addresses are used for sending to the Active Monitor, to all bridges, to LAN Manager, etc. All Token Ring NICs can handle these addresses properly, and Token Ring drivers allow software to tell the NIC which functional addresses to listen to. Because the NICs and drivers didn't support multicast, upper layers made use of these functional addresses instead. I remember that the AppleTalk developers had to go to all sorts of shenanigans to get their multicasts for zones to work right, using functional instead of multicast addresses. It appears from RFC 1469, that for IP multicast it was decided that the functional address C0-00-00-04-00-00 would be used. That's in Big Endian (non-canonical) format. Notice that it's Group/Locally administered and functional. It is not guaranteed to be uniquely used for IP multicast. Other upper layers could use it too. Anyway, that's the history. I was wondering if the problem was ever fixed. Did Token Ring NIC vendors start supporting real multicast ever, or do upper layer developers still need to use a functional address? I think that the problem was never fixed (or at least that developers don't assume that it was fixed.) The link that Frank sent, although written years ago, seems to be up-to-date and implies that IP multicast on Token Ring uses a functional address. I also found an RFC 2470 that specifies IPv6 on Token Ring. It uses functional addressees too. On the other hand, some applications may send to a multicast address that isn't also a functional address, but I haven't found any. I think to play it safe, developers assume that the Token Ring NICs understand functional addresses, but not necessarily multicast. BTW, Radia Perlman's new edition still says this too. She didn't update that discussion, probably because it's still an issue on Broken Ring. Priscilla ________________________ Priscilla Oppenheimer http://www.priscilla.com Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=45554&t=45554 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
