Re: comments on draft-py-multi6-gapi-00.txt

[Iljitsch van Beijnum <iljitsch@muada.com>]

On Tue, 4 Feb 2003, Kurt Erik Lindqvist wrote:

> You can also argue that address consumption will most likely NOT follow
> geographic boundaries. If I take Stockholm, Ericsson for example have
> their HQ in a relatively unpopulated area. Also the "swedish silicon
> valley" have a high number of multinationals that will want to
> mulithome, but relatively small population.

I'm sure address consumption won't follow the population spread.
However, there is only one other objective way of pre-allocating address
space: on geography. This both wastes a lot of address space and fails
to provide enough of it. In the end, it all comes down to people: they
own/operate equipment that needs to be connected over multiple lines.
I'm sure many more people own or run a multihomed network in the
Swedish silicon valley than in a rural province of China, but the gain
you get from allocating less for China is much smaller than the hassle
of having to manually look at it and justify your decision.

If we are generous there are 25000 multihomers at this moment, spread
over the US/Canada and Europe. That's around 600 million / 25 thousand =
1 multihomed network per 24000 inabitants. The current GAPI figures
provide for a /26 for Sweden. So that's 8833000 inhabitants / 4194304
multihomed /48s = 1 multihomed network per 2.1 inhabitants.

Obviously at these levels of multihoming the number of multinational
organizations becomes irrelevant.

Please note that although the address allocation mechanism supports
these levels of multihoming, that doesn't mean my other draft does. If
you get 4 million multihomers in Sweden and a router can only take a
250k routing table, you need to interconnect with 16 or 32 different
routers in Sweden. That's not good. (But at least this way the problem
has moved from "unsolvable" meaning loss of connectivity to "solvable by
throwing money at it" meaning we get to make an economic tradeoff.)