On 2003-08-19 at 04:14, Mark Smith wrote: > a) MAC addresses are reasonably easily to change. You can't guarantee > that the end-users will follow the recommendations ie. enable the > locally administered bit. Most networking people I've met don't know > there is a "locally administered" bit in a MAC address. > > b) There have been cases where manufacturers have allocated non-unique > MAC addresses. What is worse is that these duplications have > apparently tended to happen within the same batch of NICs, and have > been encountered when somebody goes to deploy a group of say 20 new > NICs they have just bought, and encounter one or more duplications. > > c) MAC addresses are typically placed in outgoing ethernet frame > headers by the device driver, not by the NIC itself, which is why it > changing the MAC address of an interface is usually functionally quite > easy. The device driver gets the MAC address to use from an EEPROM or > something similar from the card. Bugs in that part of the device > driver can cause duplicates.
You are forgetting: d) Some manufacturers allocate only one MAC address per host, with the expectation that two interfaces would never end up on the same network. Sun was the canonical example, but you can turn that behaviour off in the PROM. I do not know how modern Sun machines behave. /Benny -------------------------------------------------------------------- IETF IPng Working Group Mailing List IPng Home Page: http://playground.sun.com/ipng FTP archive: ftp://playground.sun.com/pub/ipng Direct all administrative requests to [EMAIL PROTECTED] --------------------------------------------------------------------