Re: [address-policy-wg] 200 customer requirements for IPv6

> How many 10K employee organizations exist in the world today (or in
> ten years)?

In the USA there are 930 such companies. This does not
include non-commercial organizations. If you count companies 
with more than 2,500 employeees, then there are 3,316 in the USA.
In the EU there are 10,000 large enterprises, that is companies
and non-profits engaged in economic activity with more than
250 employees.

I think that 10,000 employees is much too high as a cutoff
for multihoming demand. Even 2,500 is too high. So let's say
that there are 5,000 such enterprises in the USA and another
5,000 in Europe. This seems a reasonable guess. If we assume
that the combined population of the EU and the USA is 
one fifth of the world's enterprise population, then we can
expect five times 10,000 multihomers or 50,000 enterprises
worldwide to multihome.

On the face of it, that doesn't seem too bad but it misses some
things. Government organizations. Smaller companies whose desire
for multihoming does not come from the size of their business but
from the nature of their business. For instance the family website
that is running a transaction-based business that makes very slim 
margins on large numbers of transactions and therefore needs
to be up 24x7. Or a web shop that has no storefront and relies 
on a steady stream of Internet orders, maybe a small taxi firm?

This is a tough question to answer because it is entirely
non-technical and we do not have the support from economists
statisticians, and geographers to attempt a serious answer.
If RIPE and the NRO would fund some studies, we would all be
a lot better off in deciding the right policies.

> If there were consensus that the answer to the above questions was
> "yes, it's manageable", I'm sure folk would support a modified policy
> to that effect in a heartbeat.
> 
> But in the absence of some real data showing the implication of
> loosening the policy to allow the above, many (myself include) will be
> very concerned about the consequences of doing this.

The wording could be changed without loosening it up much if
we note that "customers" could refer to internal business units,
not just external customers.

--Michael Dillon