Marco Schmidt wrote:
The goal of this proposal is to reduce the IPv4 allocations made by the RIPE NCC to a /24 (currently a /22) and only to LIRs that have not received an IPv4 allocation directly from the RIPE NCC before.
The rationale for this is to make RIR-allocated ipv4 address space string out a bit longer by raising the price and dropping the size. The rationale says:
Allowing or encouraging intentional run-out of IPv4 addresses is an abrogation of the RIPE community's responsibility. Any measures that can be taken to postpone this point in time should be seen as critical.
It is not convincing to state that allocating /22s instead of /24s is an abrogation of responsibility, and as a community we should step back from this sort of overly dramatic language. A more sober viewpoint is that any future decision about tweaking the balance between allocation size and anticipated run-out time is not much more than an exercise in deck-chair rearrangement. It's a deck-chair rearrangement because the ship is going down and nothing can stop it. The proposal fails to mention the windfall that will ensue for existing address holders as the baseline RIR price for IPv4 addresses will increase from around €4.70 to nearly €19. There is no implication here that any of the authors has any self-interest in terms of proposing this sort of policy change, but there are quite a few people in this WG who would be happy with this sort of change, both brokers and incumbent address holders. Because of this, we also need to consider as a community what part self-interested financial motivation is going to play in terms of whether people are going to support this proposal or not, and how compatible this is with good governance of the policy-making mechanism for global resource allocation. The policy proposal also only skirts the issue of the cost of making this change to the Internet core. The remaining pool of 0.66 * /8 is ~11k /22s or ~43k /24s. If, as this proposal suggests, we move from a minimum routable range of /24 to /25 or longer, this is a change that comes at a cost in terms of reducing the lifetime of any routing device on the internet which takes a full dfz. On the flip side, we also need to consider how important this policy is in absolute terms. This is quantifiable because it affects 0.66 of a single /8, which is 0.3% of the entire IANA-allocated ipv4 address space. There are single early-adopter organisations out there who, if they decided to rationalise their address space, would have an effect on ipv4 depletion far in excess of any policy change that the RIPE community could make. Overall, I don't see a compelling case to make this change. Nick