> FWIW, I think NAT would have happened, IETF or no. There were people > who needed a solution, and had the money to pay for it, and there were > people who could provide the solution, and were willing to do the work. > NAT would certainly have existed. Whether it would have been widely adopted in the form that it did, had IETF acted differently is another question. Certainly a quicker development of IPng (and perhaps, development of IPng along somewhat different lines than IPv6) might have helped, but it's hard to see how that could have happened. The growth rate of IPv4 both prompted the need for IPng and hindered its standardization and deployment. Second-system effect played a role also. > It raises the question of whether there are circumstances where it's > reasonable to bend the end-to-end principle, such as when there is a > large "user" community that wants inexpensive Internet access (but is > willing to live without IP access). > One problem with the "end to end principle" is that a lot of people cite it as if it were a religious tenet (whether they're for or against it). People need to look past the "principle" and look at the reasoning that led to the "principle", along with the actual effects of the violation of that principle being considered. That reveals why the principle makes sense as a general rule and also whether violating it in some specific way is a horrible idea or only a dubious one. >>> the underlying problem was that people in the field didn't want universality >>> among endpoints, either for security or policy reasons, and people in that >>> ietf wanted universality among endpoints -- a single addressing system and >>> a single connectivity realm. that ietf said, you don't really want that, you >>> should use the internet as it was intended, and solve the problems you're >>> having in some way that preserves universality of endpoints. the field said, >>> you are completely out of your minds, we're going to ignore ietf now. then >>> later on, ietf said, if you're going to do it, then we ought to help you with >>> some standards support. >>> >>> >> That's not quite how I remember it from my POV. Some people were very >> concerned about ambiguous addressing. I don't think universal >> connectivity was as big a concern - it's not like IETF people expected >> everyone to run open networks. But mostly there was a lot of unease >> and uncertainty about NATs. Very little analysis was done. And I don't >> think that NAPTs were initially seen as the normal case. >> > > I remember such arguments. I also remember an argument that NATs were > being marketed as security devices, when in fact they did not provide > the actual level of security implied. RFC 3724 bears this out. > NAPTs were (misleadingly) marketed as security devices to consumers. But we were concerned about NATs in general, and at the time did not make the generalization that all or most NATs were NAPTs. >>> which is why i'm proposing a standard of "demonstrable immediate harm" rather >>> than the current system of "that's not how you should do it" or "that's not >>> how i would do it". >>> >>> >> That's the wrong standard, it sets the bar way too low. IETF shouldn't >> endorse anything unless it has justification to believe it is good; IETF >> should not discourage anything unless it has justification to believe it >> is bad. And that justification should come from engineering analysis >> (or measurement, if it's feasible). Sadly, a lot of people in IETF do >> not have engineering backgrounds and don't understand how to do such >> analysis. This is something we need to change in our culture. >> > > Based on some recent experiences, this type of analysis is not as > valued in the industry as it used to be. It's much more valued to be > a crack programmer; someone who can rapidly deploy something that can > be quickly brought to market. IETF cannot do its job properly if it is a slave to "the industry" and its current idea of market conditions. The Internet is not a market fad. It's a lot more like a global public works project that will have effects lasting for decades (at least) which will affect everyone on the planet. It is absolutely necessary to pay careful attention to the long term effects of its design decisions, and this cannot be done by people who feel pressure to put something on the market in six months or a year. At the same time, IETF needs to understand that optimizing for deployability first and scalability second often succeeds whereas the reverse often fails. We need to understand how to design protocols that can be deployed quickly and yet be upgraded gracefully as the real requirements for long-term widespread use of the protocols become apparent. Keith _______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf