Brian E Carpenter wrote:
> 
> For the record, I am still completely against any proposal
> that takes over the normal 16 bit subnet field, i.e.
> generates a prefix longer than /48. It just isn't
> operationally convenient.

I'm still unsure about this insistence on /48 as a critical point of
allocation.  /64 as a critical makes sense, as it is configured as the
changeover from subnet allocation to host allocation (for unicast addresses
at least).

But as for /48, consider these examples:

My home (say 4 subnets) is given a /48 for my SOHO network.
The local university (say 100 subnets) is given a /48 for the campus.
Multinational X (say 5000 subnets) is given a /48 for their entire VPN.

Only one of the above needs anything like a /48.

Next example, let's consider inside multinational X.  I want to allocate
subnets to sites.  /48 is our standard allocation, right?  But I can't
allocate /48s, because /48 is the allocation for the entire VPN, so I
instead allocate something smaller (probably a /56).

Now let's assume that said site wants to create a site-local network (not
sure why, since they're 'permanently' connected, but let's run with it). 
Why is the number /48 somehow magical for this site, considering they have a
/56 delegation?  All they really want is a way of generating internal /64
subnets.

/64 marks a well defined physical entity: the subnet, although even that can
be redefined if the address allocator wishes it.

/48 is a convenient mark for a logical entity: the 'end-user' network.  In
practice, these vary radically in size and may be further subdivided, so /48
is merely a useful convention.

-- 
Andrew White                [EMAIL PROTECTED]
--------------------------------------------------------------------
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]
--------------------------------------------------------------------

Reply via email to