Thanks, Harold, for your breakdown of how these things work in real life. With IPv6 we will be tied more than ever to a certain ISP. If I understand it, blocks cannot be used with another ISP. New ISP means new IP numbers in IPv6. The solution is simple: DNS. It was (and is) meant to move transparently from one set of IP addresses to another. From one ISP to another. It is like a PO box. Change premisses but keep your address. I think we do need to fight the charges certain money grabbers want for public IP number space. The best way to do that is to go to a competitor of those who charge. Here (Holland, Europe) it is still possible to get IP addresses for free. On Thu, 9 Sep 2004, at 15:57 [=GMT-0400], Harold Feld wrote: > At 02:52 PM 9/9/2004, Milton Mueller wrote: > >And you can't get "a single IP address" from any RIR. Perhaps you > >can from an ISP (I have never tried). The critical constraint here, > >which few people seem to understand, is the need for route > >aggregation. In other words, ISPs must be given their addresses > >in contiguous number blocks so that they can reduce the number of > >routes identified in their routing tables by lumping those contiguous > >addresses together into one route. That is why you can't have > >IP address portability under the current system. > > > As with all things, technology and policy go hand in hand. The current > policies evolved from a combination of circumstances: limitations of > technology, the need to protect legacy systems, and the concerns of the > people formulating the policy. There is, and continues to be, implications > for the real world that result from the purely technical issues of > routing. There is, and continues to be, opportunity for the considerations > of the real world to impact how technical decisions are made. It is simply > not the case that these concerns can be casually dismissed as the natural > order of things. > > The issue of noncommercial access to address space, and the impact of > address allocations on competition, are very significant issues that > deserve examination. At this point, the only thing we can recommend is > serious examination and documentation. I can only report my own anecdotal > data that a number of Community Wireless Networks (CWNs) have said they > can't afford IP address space and therefore are relying heavily on NATs -- > to the detriment of their networks. > > I believe the CWNs want to work with the RIRs and the open policy process, > but many of them don't understand how. These are often young volunteers > working to bring connectivity to poor and disenfranchised > communities. Unfortunately, the RIRs continue in the fine tradition of > believing that it is enough to have an open process and that the true > seeker, like heroes in on a quest of old, must prove their worthiness by > finding the relevant fora and figuring out on their own how to participate > (with the help of an FAQ file or two). > > The WSIS and the WGIG are broader processes. They are convened along a > more traditional public interest model that recognizes that many of the > people most effected do not have the resources to figure this stuff out on > their own and lack the capacity to come to the table and participate. > > I believe that this constituency, as the voice of noncommercial > organizations within DNS policy, has a responsibility to raise these issues > as worthy of further study. Recommendations are frankly too premature at > this stage. While I understand that there is interest in LANIC in > addressing these concerns, we have little data beyond the anecdotal data > that I have collected in a relatively short time. But because WSIS and > WGIG represent unique and appropriate fora in which to raise the visibility > of these issues, I believe this constituency should do so. > > Harold >