> Thinking a bit more about this: ethernet rules the world > right now, and I see few competitors on the horizon. Although > an ethernet switch detects attachment to a port and even > negotiates a few capabilities, there is no mechanism to > furter distribute the information learned that way (if any), > so the only way to be certain that all systems connected to > an ethernet use the same MTU is by observing the 30 year old > 1500 byte maximum, or by having an administrator police the > MTU of all attached devices. In other words: the current > situation, where jumboframe deployment is negligible.
Sticking to 1500 bytes doesn't guarantee that everything will work with Ethernet. For example, there's new stuff going on IEEE 802.3 which extends the Ethernet frame size to accommodate additional headers such as VLAN tag and security headers. The payload size does still stay at 1500. However if you have the following scenario: [new host] <-> [new bridge] <-> [old bridge] <-> [new bridge] you could easily have a situation where the 1500 bytes + new headers is greater than what the old bridge can handle. It's a bit of a corner case, but nevertheless one that we need to be aware of. Anoop -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
