Re: Draft on Globally Unique IPv6 Local Unicast Addresses

[Zefram <zefram@xxxxxxxx>]

Bob Hinden wrote:
>There is a clear tradeoff between a longer ID (to allow for better random 
>numbers or MAC addresses) and the size of the subnet field.
>
>Before revising the draft, I would prefer to hear from more people on these 
>tradeoffs.

Although I was one of those that suggested a technique that would generate
/56 prefixes, I see great value in arranging for /48 prefixes where
possible, for uniformity with RFC3177.  For randomly-generated addresses
(both centrally allocated and individually allocated), an 8-bit format
prefix and 40 bits of randomness seems like a good tradeoff.  40 bits
seems to be about the amount of entropy we're aiming at, and a handful
of bits either way makes no difference.  An 8-bit format prefix is not
too wasteful of address space.

The case for which I suggested /56 prefixes was generating a prefix
from a MAC-48.  In that case, with 46 effective bits of identifier
to fit into the prefix, we clearly can't get a /48.  It seems like a
valuable technique, but would have to be an exception to the /48 rule.
Of course, any site needing more than eight bits of subnet ID is likely
to have many MAC-48s to play with if they have any at all.

-zefram
-- 
Andrew Main (Zefram) <zefram@fysh.org>
--------------------------------------------------------------------
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 majordomo@sunroof.eng.sun.com
--------------------------------------------------------------------