At 09:38 AM 2/14/02, Michael Williams wrote: >Oleg Oz wrote: > > The originally ethernet > > chips used in cisco network modules only supported a fixed > > frame size. When ISL/802.1Q came along, the hardware needed to > > support some additional space to accomadate the VLAN info. > >I could see this as being an explanation. But doesn't Dot1Q fit into a >standard ethernet frame? I believe so, whereas ISL adds into to the >ethernet frame making it bigger. This would explain why older (or even >newer) hardware may support Dot1Q but not ISL.
Both ISL and IEEE 802.1Q can confuse older chipsets because the resulting frame can be longer than 1518 bytes. Whether these frames will confuse a chipset or not depends on the age of the chipset, and also on whether the chipset checks the length before it checks the header. ISL encapsulates an Ethernet frame, adding a 26-byte ISL header to the front of the frame, and a 4-byte CRC to the end of the frame. The first 48 bits, which would normally contain an ordinary destination address, contain a 40-bit "address," a 4-bit Type, and 4-bit User field used for priority. The first 40 bits are 0x01-00-0C-00-00. A recipient that doesn't understand ISL takes a look at those 40 bits with the next 8 bits and sees a multicast address that it doesn't understand. The recipient should just drop the packet. The exact behavior depends on the chipset, however. A good chip would just say "hey, it's not for me, get rid of it." A bad chip would say "Yikes this is way too long" and report an error, crash perhaps, or do something unpredictable. Now, the IEEE 802.1Q crowd has the advantage that they can influence IEEE specifications. They added four bytes, inserting them where a length/type field would appear normally. They were able to reserve a type field (0x8100) so that all recipients can determine if they should understand the frame or not. The first two bytes of the 802.1Q insertion are actually the Tag Protocol Identifier (TPID). The TPID is set to 0x8100. Because this number is bigger than the maximum size of an Ethernet frame, a recipient knows that the frame is not a standard 802.3 frame and that the field is not an 802.3 length field. If the recipient supports 802.1Q, it continues to process the rest of the insertion as an 802.1Q header. If the recipient does not support 802.1Q, it sees the two TPID bytes as an unsupported EtherType and should drop the frame. On the other hand, a bad chipset might barf. To support IEEE 802.1Q and avoid the need to drop and report "baby giants," the IEEE 802.3ac VLAN Tag Task Force received approval in September 1998 for extending the Ethernet maximum frame size to 1522 bytes. Priscilla ________________________ Priscilla Oppenheimer http://www.priscilla.com Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=35447&t=35245 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
