[ppml] tossed over the wall...

[Edward Lewis]

At 15:23 +0000 7/14/05, bmanning at vacation.karoshi.com wrote:
>On Thu, Jul 14, 2005 at 03:03:14PM +0000, Paul Vixie wrote:
>>  Someone else wrote:
>>  # I have to admit some confusion.  It took quite a bit of effort to rid the
>>  # routing systems of fixed network boundaries in IPv4 (and, in fact, there
>>  # are still vestiges here and there).  Is there some reason that IPv6 makes
>>  # dealing with variable length network masks more complicated than IPv4?
>>
>>  it's just easier to market/deploy something that lands on a : or . boundary.
>>  my preference would be to break that mold, once and for all, and 
>>deliberately
>>  choose something like a /55 or /57 just to force folks to THINK.
>
>	been there, done that...   the old /35 delegations were pushed back
>	to /32's... because it was "too hard"

Why make folks THINK when we have computers?

But, seriously...

Off-pretty-boundary allocations shouldn't be a challenge for 
automatons like routers.  "Shouldn't" because we hope that the vendor 
engineers can think around these things.

However, when it comes to the maintenance of the database of 
allocations (whether its the ARIN registry of some other list of 
where addresses are allocated or assigned), as well as setting up the 
DNS reverse map zones, using the pretty-boundaries makes it easier to 
manage.

If there is a reason to dole out /57's instead of /56's, then the 
reason ought to be worth the effort of having to know that "if the 
first hex digit is 8-F and there are four of them, it's in apartment 
a, otherwise it's in the house over there."

I think that its reasonable for the IETF to recommend what is needed 
- as in how much space will be sufficient for a normal household - 
but it isn't policy.

-- 
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Edward Lewis                                                +1-571-434-5468
NeuStar

If you knew what I was thinking, you'd understand what I was saying.