> I don't agree with everything in it, but djb has a pretty good > description of the problems:
> http://cr.yp.to/djbdns/ipv6mess.html Except he's got a number of things wrong. For exmaple: | The entire Internet is reachable through IPv4; only a small part of | the Internet is reachable through IPv6. The part after the semicolon is true. The part before isn't. I worked for Universitetet i Tromsø for the second half of 2002. The University arranged a house netlink to the place they had me living. This house netlink, like almost all of the radio network they had built over the island and surrounding areas, was IPv6-only; it couldn't talk directly to any IPv4-only hosts. But the entire v6 Internet, it could. This can't be the only such case - especially now, six years later. His burblage about the v6 space being parallel to, rather than an extension of, the v4 space is true as far as it goes, but is rather missing the point. He cites 2893, but handwaves it by saying that v6 proponents (as if such a large group were monolithic) say it shouldn't be used. Let's pretend for a moment that it were fully backed. The problem is, *it wouldn't help*! You'd have the same problem all over again, only with v4-mapped v6 addresses instead of native v4 addresses, because a v4-only host would be unable to talk to a v6 machine unless the v6 machine had a v4-mapped address - just as, today, a v4-only host can't talk to a v6 machine unless the v6 machine is dual-stacked with a v4 address. The only advantage would be that user software would be updated to be v6-only, with v4 handled by embedding v4 in v6, instead of user software being updated to be dual-stacked. But the software needs updating either way, so this is a distinction that makes no real difference. djb "want[s] to see a plan that, if implemented and universally deployed, will produce the magic moment". He's already described it, but pigheadedly refused to acknowledge it! The plan is simple: everyone with v4 space get at least that much v6 space, and dual-stack everything. I don't understand why he refuses to accept this plan. He seems to want a plan that not only "if implemented and universally deployed, will produce the magic moment" but which also requires no changes by most of the Internet (go read the "Excerpt from a message I sent to the ngtrans mailing list on 2002.03.20", about 57% of the way through the page). He's dreaming. There ain't no such beast. All v4-only hosts will need some changes - at the very minimum, an OS upgrade. Why does he accept some changes (OS upgardes, new application binaries) but not others (get v6 address space and connectivity)? I don't get it. (But then, that's always been true of most of what djb says.) The evolution will, I expect, be "everything is v4, with a few v6 islands" to "everything is dual-stacked" and then, as infrastructure is upgraded, v6 connectivity will gradually spread. Eventually, probably at least another decade or two, almost everything will be dual-stacked, including approximately all of the connectivity. Once enough of the net is dual-stacked that a v6-only host is useful, even if not quite as useful as a dual-addressed host, a positive feedback loop will start: more and more v6-only clients and services producing more and more pressure to convert the remaining v4-only hosts, producing less and less reason to insist on dual-addressed instead of v6-only.... djb even seems to get this, in some sense. The second-last paragraph of that page says | The way to make IPv6 addresses work is to teach every Internet | computer how to talk to IPv6 addresses---not just as an option that | the sysadmin might configure, but as something that's automatically | enabled as part of regular software/hardware upgrades. This is underway right now. I don't use Windows myself, but I've seen it said that Vista is v6-ready; modern Darwin likewise. By default. The free Unix variants already have been for years; even I, a crusty old curmudgeon still mostly frozen at a NetBSD version that's nearly seven years old, am fully v6-capable. That leaves some binary vendor OSes; many of them do v6 already, and the rest will change or die (or move into niches where connectivity to the whole Internet is not important) as the rest of the net shifts - there is enough critical mass between Windows, Macs and the free Unices that the proprietary OSes won't be able to stop the shift even if they wanted to. This will be helped along by v4 space running out, because that will start pushing up the costs of more dual-addressed hosts. As v4 space gets dearer and dearer, more and more people will decide the cost of being cut off from the v4-only world is less than the cost of getting v4 space. We're still in the very early stages. There are millions upon millions of v4-only hosts, and millions more dual-stack hosts behind v4-only connectivity. But as equipment - at connectivity providers, at end users, at service providers - is upgraded, the v6-capable Internet will start to jell. It's already starting; for months now, there's a mailing list, whose mail comes to me from a bell.ca smarthost, which I've been seeing using v6. For a provider as big as Bell to be using v6 like that is extremely encouraging to me. /~\ The ASCII Mouse \ / Ribbon Campaign X Against HTML [EMAIL PROTECTED] / \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B _______________________________________________ timekeepers mailing list [email protected] https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
