IBM Token Ring networks are
nearly identical with the IEEE 802.5 specifications that were based on Token Ring. 802.5 does not specify a topology; IBM specifies a star
topology using a multi-station access unit. 802.5 does not specify a medium; IBM specifies twisted-pair wiring.
issues with 802
- Each LAN committee can arbitrarily define addressing schemes (and multicast addresses, with 802.5, at least). While the addressing is alike for the most part, there is this bit-rearranging foo that occurs when bits are transmitted. 802.5 had heirarchical addressing to begin with, at 6 bytes comprised of the ring number and the station number. That was changed to 6 station address bytes. There are two bits at the end of the packet that are used as destination acknowledgment: the A bit and the C bit. A is a confirmation of the address, and C as confirmation that the packet was copied into the buffer. They also help determine the physical order of stations in the ring. The most significant bit is transmitted first in 802.5 (and FDDI), so bridges must rearrange the fields of the address when forwarding to a different LAN type.
- The 802.5 chips don't support multicast addresses; rather, they support 31 functional addresses (multicast addresses number into the trillions. ahem.) So, a bridge between an 802.5 LAN and any other type of LAN must be backwards compatible for address conversion.
Also, a phase lock loop is unable to lock with inevitable accumulated clock jitter at each station, but a bridge can remedy this to an extent.