Re: [OPSAWG] Heads up - Plea for allocating a /8 to ISPs

________________________________

From: [email protected] [mailto:[email protected]] On Behalf Of
Christopher LILJENSTOLPE
Sent: Friday, November 12, 2010 9:46 AM

The risk is that many of the authors then go and suck up much more than the
/10 to do the same thing, using globally unique space, when it really
doesn't need to be globally unique.  It's wasteful, and the authors are
trying to conserve as much as possible, while not blanking a large amount of
heritage IPv4 space.  Many of us were hoping to get to DSLite, NAT464, IVI
before this happened, but we are out of time, and the infrastructure (in
this case CPE) isn't ready at scale in a format that most of us can consume
and put into production in the remaining time.

[Wes]
Chris, you haven't given any sound reasoning as to why you are going to
suddenly need a significant amount of additional IPv4 space that makes me
believe the concept "if you tell us no, we'll just go get a bunch more
routable space and empty the table" is anything but a rhetorical flourish.
1) you have enough space for your current customer base, or else you'd
already be in this situation and have figured out a solution *
2) if you seriously expect your customer base to grow *dramatically* (eg
enough to justify a large block of global space at an RIR), why are you
assuming that it will be with old CPE? Churn from one provider to another
where people bring their old kit isn't likely to drive you to need 4M
addresses. Some providers already have policies that won't let me do that if
I wanted to...

You also need to prove why using global space for this is better than having
it be available for the myriad uses that *require* global space (like the
external NAT pool for all of these NAT444 and NAT64 implementations).

*As I said, the only exception that I can think of to this would be mobile
networks where they are doing significant oversubscription of their DHCP
pools (eg servicing 50M users with more like 10-15M addresses), and they
don't have the same CPE problem you're claiming prevents you from using 1918
space.

Wes George

_______________________________________________
v6ops mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/v6ops