[ppml] Proposed Policy: IPv6 HD ratio
bicknell at ufp.org
Tue May 10 16:57:41 EDT 2005
In a message written on Wed, May 11, 2005 at 06:18:45AM +1000, Geoff Huston wrote:
> I would encourage you to have a look through the presentation
> for a background to this area of consideration.
I find myself twisted up thinking about this, so let's see if I can
explain myself clearly.
I believe the current HD Ratio implies a linear relationship between
network size and levels of hierarchy. Specifically, every 3 bits
(factor of 8) you get one more level of hierarchy. Changing the
HD ratio changes the number of bits to get one more level, but does
not change the linear relationship.
If we start with a /32 having a base of "N" layers of hierarchy,
we have potentially 10 layers more for someone with a /2. We don't
know what N is for any particular provider (and indeed, it will
vary from provider to provider), but I suspect N for most providers
today is somewhere on the order of 1-6 layers. This gives us a
total range of 1-16 layers.
The problem is, 16 layers seems like a bit much. Even being generous
with a progression like World -> Continent -> Country -> Region ->
State -> County -> City -> MSA -> MSO -> Customer we're at only 10
layers, and most people aren't going to divide the address space
at all of those boundaries. Also, as the number of layers increase,
the lack of efficiency in the smallest ones make less difference. A
provider with a /20 will have no issues with an MSO not having a
single customer in it.
I think the growth of hierarchy is what grows logarithmically. The
more layers you add, the slower you add layers. Going from a /32
to a /30 you might add one layer, but the next layer will take you
to a /26, and the next layer to a /18. I'm not sure how you turn
a logarithmic growth in hierarchy into a formula that makes sense
At the same time, I wonder if it's overly complex. If we're going
to require the biggest people to have ~50% efficiency, why not just
require that of everyone? Why are we holding the smaller people
to a tighter standard that they will simply be able to relax with
size? Philosophically it doesn't make a lot of sense to me.
Leo Bicknell - bicknell at ufp.org - CCIE 3440
PGP keys at http://www.ufp.org/~bicknell/
Read TMBG List - tmbg-list-request at tmbg.org, www.tmbg.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 187 bytes
Desc: not available
More information about the ARIN-PPML