I think Wittenhorst proposed pricingmodel may be a good idea, but there are also several issues with it (as with the current pricing scheme). Our LIR would have to pay twice the amount that we are currently paying today with Wittenhorsts suggestion. Current price: €1800 New price: €3432 (2x /19's) Perhaps a pricing model shouldn't be based on prefixes but instead on the amount of ip-addresses in a non-linear model. IPv6 addresses and ASNs too. Paying for an "object" in a "database" is kindof silly as there are little or no expenses related to a database itself these days. If it would be up to me I would get 1x /18 instead of 2x /19 in the beginning, but RIPE NCC didn't approve on that. I shouldn't be punished because I am "saving" resources by using /19s instead of a /18 which was forced upon me by RIPE NCC. The amount of "objects" a prefix require in a proprietary database may also be subject to change at any time. On 08/03/11 11:23, James Blessing wrote:
On 02/08/2011 22:14, Marcel Edler (Optimate-Server.de) wrote:
In new charging scheme I have to pay 5000 Eur for 73.440IPs (1x/16,1x/19) like an ISP with 1.024.000 IPs (/12)! I pay 0,068 Euro for each IP. The ISP with 1.024.000IPs pays only 0,00488 Euro for each IP. I have to pay !!14!! times more than the other ISP!
Er, what about the other db resources? ASNs, IPv6 space, PI etc? If you want a per 'thing' pricing structure would it not be better to charge:
Membership X,000 PA (or X00) db object Y PA (i.e. each object in the db be that a /24 or /8 of v4 space, a /32 or an ASN)
This would then better reflect the 'impact' on the db (it might also encourage a better maintenance of the db) you could of course charge different value for each type of object if the NCC can quantify the 'impact cost' of each type of resource.
J