Re: [v6ops] PI [ULA draft revision #2 Regarding isolated networks]

[Tore Anderson <[e-mail]>]

* t.petch

> The damage done by PI in IPv4 was understood before the RIRs
> started handing out addresses in large numbers, so while PI exists,
> it has always been hard to get.
> 
> The fear with IPv6 is that just because one constraint on PI has
> been removed, those handing out addresses will not realise that there
> is another show-stopping constraint in the number of entries a FIB
> can cope with, 1M being the best estimate (as before, on the RRG
> list) with the foreseeable improvements to current technology.
> 
> And I think that every SME who has lost business with the
> unreliability of their ISP will want multi-homing and will think that
> with IPv6 and PI the constraints have gone, and the number of such
> SMEs can only approach 10M over time.

I'm only familiar with the RIPE region's policies, but PI certainly
wasn't «hard to get» for an SME here that wanted to multi-home (until
the point where we stopped completely issuing new blocks of it, that
is). It only had to do the following:

1) Pick a sponsoring LIR out of the ~10k available to choose from, and
2) Fill out the appropriate request form, and
3) Be prepared to pay a yearly fee of €50 (+ the LIRs profit margin).

The procedure was the same for IPv4 PI as it still is for IPv6 PI.

(The only real difference between IPv4 and IPv6 PI was the difficulty in
getting a large IPv4 assignment. While in IPv6 you get an enormous /48
with no questions asked, in IPv4 you only get a measly /24 and would
have to provide additional documentation if you wanted more. However
this difference is moot in a discussion of FIB sizes, as the scarce
resource is the routing table entries themselves, and the prefix length
of each individual entry is AFAIK irrelevant.)

Tore