back to article Small carriers aren't showing up to IPv6 standards chats, consultant warns

Smaller ISPs are dealing themselves out of discussions about the inevitable transition to IPv6, a Spanish consultant warns, and could find their future defined by large telcos. Frustrated at their indifference, Jordi Palet Martinez of Consulintel has appealed for just a bit more enthusiasm (and participation) from ISPs in IPv6 …

  1. Anonymous Coward
    Anonymous Coward

    Ugh.... HomeNet ...

    Out of all IETF delusions, this one is probably the worst. IT exemplifies how the IETF has lots its way and is no longer the standard organization from the days of John Postel.

    HNCP and HOMENET is a pile of demented drivel pushed by the few remaining vendors interested in LARGE home CPEs, designed around geeky use cases (I have N links to my house, I need redundant routing for that) and using technology which no Joe normal user can even contemplate debugging.

    I attended a couple of the early HOMENET meetings and left before the end of the session as whatever the people in the meeting were smoking they were not keen on sharing it.

    I am all with the small ISPs. The consluttant promoting this drivel can get lost. While TR69 is the spawn of the devil it is a well known spawn of the devil and it works.

    The large carriers "showing up" are also people who have nothing to do with actual network deployment. They are "standards junket" attendees whose job in a large carrier is just that - to be a zombie present at all possible junkets and have a "standards engineer" as their title.

    1. Yes Me Silver badge

      Re: Ugh.... HomeNet ...

      Homenet or not is nothing to do with large or small ISPs. It's to do with simple or complex networks in the home. Once the ISP has delegated an IPv6 prefix to the home gateway, the ISP has nothing to do with the internal complexity (or simplicity) of the home network. And you've missed the point of the IETF homenet work: it's to make home networks 'plug and play' whether they are simple or complex. Whether a particular home gateway vendor does or doesn't include homenet support is a commercial decision, of course. What Jordi is worried about is whether RFCs that end up in procurement specs serve the interests of all ISPs, or only some. That's a quite different question from whether ordinary citizens will end up with complex (multi-router) networks in their home or small office. That's a commercial decision too.

  2. Kevin McMurtrie Silver badge

    Not in the timeline

    I've always tried to use small ISPs because they're generally more likely to give you an honest internet connection and honest service. Unfortunately, they don't live long. They start off really good at fine tuning the system they have but they never learn to scale, acquire investment money, or upgrade. Eventually they're stuck with a reputation as the "best dialup" or "best DSL" provider years after those products are obsolete. Maybe they limp along as a re-seller for a few more years. If they have no plan for the future, there's no way they're going through all the hassles and investments to get IPv6 running.

    It's a shame because small ISPs usually don't know what a mess the competition is. AT&T is still mailing me advertisements for their 20 year old DSL1 garbage. AT&T can't even support incoming connections because they're still using an on-demand PPP layer.

  3. bazza Silver badge

    It is interesting how IPv4 is exhausted, and yet here we all are quite happily on the net... I know that most mobile networks are v6, perhaps that's why we've managed this far.

    There is kind of a conspiracy though. IPv6 does or should make it trivial to connect to devices in the home from outside. That ought to make it easy to have IoT devices that are directly controllable from one's mobe, and no need to register with MegaCorp's server (ok I'm asking you all to suspend disbelief and imagine that IoT is going to be a massive thing...).

    The thing is that the current IPv4 arrangements, namely using MegaCorp's server as a broker between IoT device and mobe, suits MegaCorp's very, very nicely indeed; they can slurp the data.

    1. Voland's right hand Silver badge

      There are fundamental technical reasons for it

      Looking around, good, no v6 geeks or $deities in the vicinity so I am not going to look like one of those Looney Tunes "flattened character" animations after being run over by an angry rhinoceros (let's say Fred Baker or Mark Townsley).

      The fundamental technical reasons for this are as follows:

      1. Most of the problems of v4 in the initial problem statement for v6 have been solved. Long ago. One way or another. v6 either breaks the solution, has incomplete and complex support requiring multiple network elements or does outright incorrect and idiotic network design assumptions which have never been fixed. Example - device configuration. v6 erroneously hands out power for this to the router(s) via router advertisements, cannot be fully configured without multiple solutions present (with accompanying complexity) and all of this has been solved long ago in DHCP. The whole v6 autoconfig in this day and age is a solution looking for a problem. Instead of admitting it, revising the protocol and obsoleting all the b***cks in it, the v6 geeks deliberately crippled DHCP so you cannot get v6 config in the v4 reply and v6 version is subtly incompatible with v4.

      2. The end to end principle is idiotic in the highest order. It was a jolly good idea when the machines on the Internet were in their thousands, each had an admin and there was some responsibility involved. It started looking a bit nuts when the machine count went into the millions with virus ridden 95s joining the net. It is a complete lunacy when the machines are in the billions and more than half of that are embedded builds which get no patching. YOU DO NOT WANT THESE E2E accessible.

      3. The address exhaustion problem has mostly gone away as a result of developers switching from protocols which were NAT unfriendly to protocols which do not give a damn about how many NATs they traverse.

      No v6 $deity will come even close to admitting any of these. They start any of their talks with "v4 is obsolete". They have made their careers on making v4 and v6 incompatible and pushing v6 at any possible occasion (v4/v6 DHCP is a prime example). You are more likely to make them commit sepuku with a blunt kitchen knife than admit that v6 NEEDS a major revision with half of the features obsoleted straight away.

      1. Anonymous Coward
        Anonymous Coward

        Re: There are fundamental technical reasons for it

        Hear, hear.

        +1000

      2. Blitheringeejit
        Thumb Up

        @V'sRH: +lots, and furthermore...

        The IP4 address scarcity would be much eased if large North American address space users (DoD, colleges, gov etc, not to mention large corporations) would desist from giving every machine on their LAN a public IP address. This could free up massive amounts of address space for the more recent arrivals to the table.

        And NAT doesn't just solve the address scarcity problem, it also protects (to some degree) against direct port hacking, and is generally a Good Thing - especially given your well-made point 3.

        Disclaimer - I understand v4, but try as I might for a decade or so, I've been unable to get my head around the basic mechanisms of v6. I don't claim to be the sharpest tool in the box, but it does feel like an over-complicated solution to problems which don't need to exist in the first place. ICANN just needs to get tough with IP4-hoggers.

        1. Len

          Re: @V'sRH: +lots, and furthermore...

          If you do the maths on the unused IPv4 address space you'll find out it is a pointless exercise. It would require a bunch of large organisations and multinationals to overhaul their existing network infrastructure. An operation that would not be quick or cheap.

          If you compare the amount that could potentially be freed against the IPv4 depletion rate you'll find that you have spent many millions and a couple of years to free up between two and three month's worth of IPv4 addresses.

        2. Anonymous Coward
          Anonymous Coward

          Re: @V'sRH: +lots, and furthermore...

          The IP4 address scarcity would be much eased if large North American address space users (DoD, colleges, gov etc, not to mention large corporations) would desist from giving every machine on their LAN a public IP address.

          Sadly, this is true - I used to work at a major University that has TWO ENTIRE Class B subnets, one of which is allocated to, and controlled solely by, the CS Department. That was probably a good idea in 1985, but not so much anymore. When I left three years ago, every single computer still had a directly routable IP address.

      3. Christian Berger

        Re: There are fundamental technical reasons for it

        Yeah, so you're saying that I should use G-Mail instead of my own mailserver?

        And seriously, with UPnP all those crappy devices you don't want to have on the net are _already_ on the net, despite of NAT, they simply open port forwarding.

        We have lots of problems already which stem from NAT and which aren't addressed by advances in IPv4. Look at mobile devices. People cannot even transfer a file from computer A to computer B because they are likely in 2 different NATs. That's why we even have data-slurping companies like Instagram or WhatsApp. Nobody would even think about using that if there was E2E communications, you'd just use the network like you use the telephone network.

        "Smart" devices often have to go through a central computer so you can talk to them through the net. If that vendor decides to shut down that service, you can no longer use that feature.

        IPv4 simply separates the net into 2 classes of users. Those with public IP addresses, and those without. Increasingly people, particularly in poorer countries, end up behind multiple layers of NAT.

        1. Blitheringeejit
          Stop

          Re: There are fundamental technical reasons for it

          If you have your own mailserver, presumably you also have the smarts to program up a router with port-forwarding to make it work..?

          And meanwhile, do you really need your fridge to have direct public address access to my fridge? That's a surefire recipe for a Terminator scenario...

        2. Voland's right hand Silver badge

          Re: There are fundamental technical reasons for it

          Look at mobile devices. People cannot even transfer a file from computer A to computer B because they are likely in 2 different NATs.

          If those are MOBILE devices, can we all ask it stays that way. Please? Pretty please?

          E2E is a lovely idea in principle, in practice, specifically for mobile where 90% of the devices are unpatched Android with at least half of them spouting a DLNA implementation with a buffer overrun it better stay behind NAT. While NAT is by no means perfect, NAT and the deliberate filters at SP which prevent one mobile talking to each other is what is saving us from zero day worms exploiting all of these.

      4. Len

        Re: There are fundamental technical reasons for it

        Your problem no.2 has been solved quite some time ago with the invention of firewalls.

        Whether IPv4 or IPv6, nobody is suggesting to disable the firewall in the CPE. As such, end to end networking over IPv6 is only possible if the firewall is configured to allow it. I have yet to find an IPv6 compatible CPE that by default allows incoming traffic without the connection being initiated from inside.

        1. Voland's right hand Silver badge

          Re: There are fundamental technical reasons for it

          Your problem no.2 has been solved

          Correct. However, when you listen to the high priests of v6 preaching us the goodness of the v6 e2e principle you will never hear the word firewall. It is not part of the holy e2e paradise. The fact that some of them also have high security clearance and consult 3 letter agencies should not surprise you either.

          In any case - there is nothing wrong with v6 purely for address depletion purposes if it is obsoleted as it is today, all of the IP address autoconfig crap removed while fixing DHCP so that v6 config can be dished over v4. Some revision of the flow label semantics so they are usable would also be nice but not essential.

          The problem is that the v6 fanatics insist that there is nothing wrong with the crazy mix of v6 autoconfig + v6 DHCP as well as crippling the v6/v4 DHCP interop. They perfectly understand that dealing with that is one of the "v6 way or the highway" moments for an organization and they refuse to budge on it to achieve a personal goal - promotion of v6. At everybody's else expense. Obsoleting this aspect and doing an errata on all relevant v6 RFCs is so overdue, we might as well do v7.

          Ditto for the v6 aspects and the mandatory v6 support in HNCP and HOMENET (where this thread started).

      5. Fred Goldstein

        Re: There are fundamental technical reasons for it

        Amen. Friends don't let friends use IPv6, the Children's Crusade Protocol.

        IPv6 is one of those Milgram-style psychological tests, wherein people are told that some "authority" said to do something, so they set aside their own judgment and assume that the authority is right. If those young people trained in the TCP/IP monoculture only knew the history and background of IPv6, they might begin to question the authority.

      6. Yes Me Silver badge

        Re: There are fundamental technical reasons for it

        1. " The whole v6 autoconfig in this day and age is a solution looking for a problem." If you're running a reasonably large enterprise network, DHCPv6 is probably useful with today's old-fashioned (basically manual) method of network design and config. That will change over the next 10 or 20 years - you're going to see much more automated provisioning and much less human decision making about trivia like addressing schemes and prefix assignments. The end game is full automation, of which IPv6 stateless address autoconfiguration will be a small part.

        2. "The end to end principle is idiotic in the highest order." You don't understand it. It doesn't mean any host can talk to any host. It describes how to design protocols for two hosts that can talk to each other. Different thing.

        3. "developers switching from protocols which were NAT unfriendly to protocols which do not give a damn about how many NATs " It's just sad that you believe that is a good thing. It's a quick fix, that's all.

      7. Anonymous Coward
        Anonymous Coward

        Re: There are fundamental technical reasons for it

        The v6 thing I absolutely agree on. The "no E2E" is ridiculous. Yes, I *DO* want E2E, everywhere, all the time. The *DEFAULT* should be to block all traffic at the firewall. NAT doesn't fix that problem, the firewall does. I can just as easily misconfigure UPNP and v4/NAT solves nothing. The fact I *CAN'T* get E2E without ugly hacks is a problem, and it fundamentally breaks the internet.

        Saying that carrier grade NAT fixes the address exhaustion problem is like saying the polar ice caps fix the water shortage issue in Africa. Just because it's a fix, doesn't meant it's a band-aid covering up an even bigger issue down the road. Oh, right, I can just rely on some third-party intermediary connecting two devices when they're both NAT'd five layers deep. Of course those servers aren't free, so suddenly *EVERYTHING* on the internet is a paid service or selling your personal information in order to provide said service.

    2. Yes Me Silver badge

      An interesting conspiracy theory

      " IPv6 does or should make it trivial to connect to devices in the home from outside. " Well it doesn't if you have a firewall, which every IPv6 CPE does as far as I know. And it shouldn't. Actually it needs to be hard to connect from outside, so that only the authorised parties can do so. But you are correct that IPv6 is an enabling technology for things that we really can't do in a straightforward way with one address and a NAT box for a whole home network. So there is a conspiracy to make things better.

      I think Jordi's a bit pessimistic though. I've seen many small ISPs offering a good IPv6 service well ahead of the dinosaurs.

      1. jordipalet

        Re: An interesting conspiracy theory

        Yes, but the difference is ... up to now we have done dual-stack deployments. CPEs support that since a long time ago.

        Now we have two options, using CGN (which is not good, gets blocked very soon by service providers such as Sony PSP), or using the latest transition tools such as 464XLAT. And here we have the problem, is not required in the standards, so small and medium ISPs don't have access to that. A big ISP, which purchases big amounts of CPEs in every deal, they can ask the vendor to do it just for them, but this doesn't go to the market ...

  4. John Sager

    Religion on both sides

    There are religious 'keep V4'ers just as there are religious v6ers. The address problem isn't going to go away, NAT or no, and v6 is the only game in town to address that. If its protocol support is too geeky for some, then essentially all that should be hidden away in the CPE. Unfortunately most CPE makers are producing crap products with features you don't want (UPNP) and a complete lack of features that are now quite important (QoS). Apart from one or two, I'm not sanguine about the prospect of getting performant v6 kit anytime soon. I went down the 'roll your own' route to fix this a long time ago but then I'm in a small minority of punters that can.

    1. Anonymous Coward
      Anonymous Coward

      Re: Religion on both sides

      There are religious 'keep V4'ers just as there are religious v6ers

      Very true. Regardless of what anyone thinks, IPv4 is now in the "the only debate is how long" phase of dying. NAT is a right PITA for many reasons, and CG NAT even more so. Some parts of the world no longer have enough IPv4 addresses to allow individual users to have a single IP per subscription - hence many people have to sit behind CG NAT.

      Even if CG NAT didn't break anything (which is does), it's expensive - so a cost to carriers which ultimately filters through to the users.

      Unfortunately most CPE makers are producing crap products

      I'll single out the complete pile of manure shipped by Vodafone as the epitome of that. It's got a user interface that seems to be aimed at 2 year olds, that just doesn't do some simple things, makes other basics way too complicated - and worst of all if you ever have the misfortune to try and connect to anything while the DSL line is down then it'll hijack all your connections, cause a flurry of security warnings, and generally throw a pile of manure in your browser cache.

      I partly agree with Voland's right hand above where he says The problem is that the v6 fanatics insist that there is nothing wrong with the crazy .... Watching the threads in the IPv6 ops mailing list, it's clear that there are some entrenched positions - and definitely some "what works for us is X, Y doesn't work for us, therefore no-one should be able to do Y" approaches. I couldn't possibly suggest a link between that and the fact that Android does not support DHCPv6.

      Routing information is another area where (from my viewpoint of operating small networks) there is a problem. The argument FOR Router Advertisements is that the router guys are a different team to the DHCP guys (which they probably are in very big networks who are disproportionately involved) - hence allowing the router guys to configure device routing is the right way as it means they don't have to co-ordinate with the Server & DHCP guys. Dunno about you, but I can't think of a situation where the network guys can alter the addressing scheme without co-ordinating with the Server & DHCP guys (or vice-versa) so that's a non-argument. But what it does do is push routing decisions to end devices - so instead of being able to put your routing rules in one (controlled) place, you have to use a currently non-existent protocol to push that down to all your clients and hope that they all co-operate !

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon