On 12/3/2011 5:54 PM, Henning Schulzrinne wrote: > Almost all residential customers will use a standard home router; as > long as that home router does not make the new space available to > customers, it will not be used. Almost all residential users get > their home NAT box either from the ISP (who obviously won't ship such > a box) or from one of a handful of retail consumer equipment vendors, > who won't suddenly switch from RFC 1918 addresses, either (because > they don't want to get the support calls). I think you're right about that, however it's also true that almost all of those devices use 192.168.[01]/24. So for that market you can easily avoid the problem on the CGN side by using a different 1918 block. > I don't think your consumer ISP will have much sympathy if you call > them up and tell them that you decided to use 128.59.x.x internally, > reconfigured the gateway and can no longer get to Columbia > University. Apples and cumquats. "Hijacking space that is allocated on the public Internet" is an entirely different situation than "Using space that's not supposed to be public for a different purpose than it was intended for." Yes, I realize that both are technically "wrong" in the sense that they are against the rules, but I can easily predict the arguments of the sysadmin responsible for the latter ... "We never thought we'd be behind a CGN, so we figured it was Ok to do it." "Our old provider didn't use that range for CGN, so we thought it would be Ok if we did." Etc. > This is an economics issue: If one big corporate customer with a > too-creative sysadmin calls up after "finding" this new address > space, this can be dealt with. (Indeed, that large corporate > customer probably has non-1918 outward-facing addresses to begin with > and will keep them, so they are the least likely target of CGNs.) If > 10,000 consumer customers call up because their Intertubes aren't > working, the ISP has a problem. > > Thus, I'm having a hard time believing in the theory that the new > space will be immediately appropriated for consumer ISPs. By whom, > exactly, and on what scale and with what motivation? Actually I think your analysis is 100% accurate, it's just your conclusions that are faulty. :) This is a 90/10 problem. 90% of the problem can be addressed by simply using a 1918 block that is not one of 192.168.[01]/24. (Where 90% is a WAG of course, but you get the idea.) However, the enterprises that are the most likely to have exhausted, or nearly exhausted 1918 space already are, for that exact reason, the same ones who are most likely to (mis-)appropriate the new space for their own 1918'ish needs. So the vast majority of the problem can be solved easily with 1918 space. The rest of the problem is incredibly likely to cause pain no matter what block is chosen for CGN space. So, there's no point to doing the allocation. Doug -- "We could put the whole Internet into a book." "Too practical." Breadth of IT experience, and depth of knowledge in the DNS. Yours for the right price. :) http://SupersetSolutions.com/ _______________________________________________ Ietf mailing list Ietf@xxxxxxxx https://www.ietf.org/mailman/listinfo/ietf