On 2 Jun 2026, at 10:20, Chano Klinck Andersen wrote:
> It’s interesting how often the assumption is made that IPv4
> addresses will be returned if they cost more to hold on to.
Well, I don’t see how such a change would make it worse than what we
see today?
For a lot of organizations, the ~€2-3k they have to pay for their
multiple /16’s or whatever that they don’t really need, is a
rounding error in a lot of budgets (and “a couple of /16’s is being
generous — some have WAY more that they barely use).
But for a small ISP with a /24 or whatever, the ~€2-3k+++ is starting
to actually matter in their total budget (I know because I’m one of
them).
> The way I see it, they will not be returned. Instead, they will be
> sold/leased to the highest bidder. Higher fees to RIPE NCC will make
> IPv4 addresses more valuable as an asset, which in turn would raise
> the market price.
That assumption only applies to those who are already leasing out excess
address space they own… and they might also re-consider if their fee
to RIPE would increase significantly. That would be the same logic as
landlords in some countries that are selling their rental properties due
to increased loan interest (aka its not profitable anymore).
I don’t know if you still don’t have to pay for legacy address
space, but that used to be the case. That’s literally zero incentive
to do anything but “sit on it” (unless you speculate to sell it off,
as many also have done). And those organizations that do register, can
get away with the equivalent of PA-as-PI address space via sponsoring
LIR (AFAIK) with essentially zero cost as well (compared to the current
price of being a LIR).
Ironically, for those ISPs that have “a lot of resources”, the
“per resource”-model for sponsoring LIR would be more fair than the
current flat-fee (i.e. they would have to pay more than they currently
do via the flat-fee).
I also know of individuals with /16 as PI via sponsored LIR. They pay
like €150 or whatever, and use it for “shits & giggles” (sure they
use *some* part of it for labbing or whatever, but still).
Individuals/organizations that have these amounts of resources that are
not strictly needed would for sure either a) sell, or b) return to the
pool, if they had to pay “accordingly” based on their resource size.
> Instead of trying to reclaim IPv4 addresses (which I don’t believe
> is going to happen in any meaningful capacity), what about trying to
> figure out how we can complete the transition to IPv6, making IPv4
> addresses irrelevant and ending this yearly recurring discussion once
> and for all?
Even if we do at some point, the dual-stack scenario is still going to
be there for many decades still. Sure, we’re at a better place than
just a few years back, but it’s going to take time. A long time.
> The only reason, as I see it, why some members want the tiered or fee
> per IPv4 address model is because they can’t get the IPv4 addresses
> they need. If we had all just transitioned to IPv6 a long time ago,
> when the problem became apparent and before the last /8 was used up,
> no one would be complaining about the charging scheme the way they are
> today.
I simply want it because it’s fair to pay more for more things. If you
want 100 cars, you have to pay for 100 cars.
As a consultant I have worked for more organizations than I can count
that has way more IPv4-resources than they would ever need. And they
just sit on it because it doesn’t cost them anything.
Some of them are universities that historically had a flat, one-VRF
network, zero firewall, and “fully open for everyone due to being
research oriented” with public IPs on every device. That was fine
historically, and I don’t blame them. However, most of these have now
transitioned to fully segregated networks with firewall and NAT-boxes
due to the increased need for cyber security.
As part of that change, many of them started to use RFC1918 address
space. The old argument that “we use our public addresses because then
we don’t need to have expensive NAT boxes” are no longer valid.
But do they do anything about the excess IPv4 address space they have?
No, they just keep it because it’s essentially zero-cost for them,
even if they literally don’t use it anymore (or maybe they use a
couple of /24’s of their multiple /16’s as NAT pool).
Should they return it to adhere to RIPE policies? Of course. But
there’s no way to reliably check if they actually use it or not. Which
is part of the problem.
> Prior to IPv4 run-out, we rarely ever discussed the fees, but since
> running out, we have the same discussion, with the same arguments, and
> the same result (no change at all) every year.
I don’t technically care about the lack of IPv4 address space, and the
need to return it. Its just a nice way to solve several problems;
* Make sure that you pay for the service you get according to the
size/value (ASN/IPv4/IPv6-resources)
* Give good monetary incentive to return unneeded resources to RIPE
(or sell it, both solves the problem that they don’t have to pay for
it anymore)
* Reduce the cost for smaller LIRs and increase it for the bigger LIRs
(pay more for more things)
> Furthermore, RIPE NCC is a membership association. In many
> associations, members pay the same fee even if they make different
> levels of use of the shared facilities or services. I have rarely seen
> associations require members to pay based on their wealth or exact
> usage of shared resources. I’ve seen differentiated prices for kids
> vs. adults, but that’s about it.
If we only received equal service (whois, etc), then I would agree with
your argument. But that membership also gives you access to a resource
that has a real-life monetary value, and the more you have of it, the
more valuable is it. There is no way around that fact, no matter what
you argue.
> The discussion on RIPE NCC providing more than just barebones registry
> services, the way I see it, is really a question of “for the good of
> the internet community as a whole” vs. “for the good of my bottom
> line”.
I agree (hence why I said I like the idea of a lot of these extra
services). But they still benefit a big international ISP more than a
small regional one.
And the big international ISP usually has a lot more money, so having to
pay 10x or even 100x the current LIR fee would in most cases be
unproblematic (in some cases just a rounding error on their budget).
I’m not saying “pay more for more resources” will be without
problems. But I think it will be a LOT better than the current price
model.
--
Joachim
