Obviously it doesn't matter, but providing an explicit procedure to generate the 'good enough' unique number that doesn't depend on one of the EUI-48 values embedded in a device, will eliminate questions about which is the best EUI-48 address to use from IPv6 mailing lists / discussion forums, support lines etc, and possibly this mail list, even when there is no best EUI-48 to use.
Doesn't exactly the same issue present itself with birthdays? eg: Which is the best birthday to pick? Mine or a descendents? And all the unix guys will pick 1/1/1970.. ;-)
Having a procedure that can avoid asking the user to type anything in at all (ie one that reads and EUI from a device) sounds like a good thing to me.
Using an EUI-48 also doubles the number of "unique" bits that are input to the hash algorithm. Birthdays are rounded to the day, regardless of whether we represent them in 64bit NTP format.
The current time of day counter doesn't seem to me like it will generate all that many unique bits either -- two (or more) routers configured around the same time will have a whole lot of bits in common.
- aidan
-------------------------------------------------------------------- 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] --------------------------------------------------------------------