OMB: IPv6 by June 2008

Jay R. Ashworth wrote:

Well, with all due respect, of *course* there isn't any 'killer site'
that is v6 only yet: the only motivation to do so at the moment, given
the proportion of v4 to v6 end-users, is *specifically* to drive v4 to
v6 conversion at the end-user level.

We need either one efficient v6 p2p application or sites providing free p0rn only over ipv6 connections.
The same would work for multicast.

Pete

Huh, Europe is moving to IPv6? I must have been asleep at all industry
meeting in the past few month and years...

  From what I've seen at the RIPE meetings, Europe is definitely moving towards IPv6. Maybe not as fast as some parts in Asia, but it's definitely moving that way. Moreover, it's moving towards IPv6 much faster than the US.

                All the other stuff and the different address scopes are
not only impractical but confuse the average consumer and MCSE admin to
no end (and those are the people that have to deal with it all the time).

  IPv6 has its problems, yes. There are implementation issues that confuse programmers at Sun working on Solaris, and confuse network application programmers with a hell of a lot of experience under their belt. If you can't talk directly to Jinmei himself, you're likely to be well and truly screwed.

  But just because IPv6 has problems doesn't mean that it doesn't solve the fundamental address space problem in IPv4, and doesn't mean that it is anything less than the best available alternative.

Brad Knowles wrote:

Huh, Europe is moving to IPv6? I must have been asleep at all industry
meeting in the past few month and years...

    From what I've seen at the RIPE meetings, Europe is definitely moving towards IPv6. Maybe not as fast as some parts in Asia, but it's definitely moving that way. Moreover, it's moving towards IPv6 much faster than the US.

I don't care what you see at RIPE meetings. You have to look at how
many servers/services are reachable via IPv6. Nothing else. Sure,
many European ISPs have got their IPv6 prefix and some even announce
it via BGP, but actually using it for anything more useful than "hey,
I can ping6 you!" is far off.

                All the other stuff and the different address scopes are
not only impractical but confuse the average consumer and MCSE admin to
no end (and those are the people that have to deal with it all the time).

    IPv6 has its problems, yes. There are implementation issues that confuse programmers at Sun working on Solaris, and confuse network application programmers with a hell of a lot of experience under their belt. If you can't talk directly to Jinmei himself, you're likely to be well and truly screwed.

Ain't this *the* problem??? If not even Joe OperatingSystemEngineer
can understand it, what is John Doe at home supposed to do?

You know, there is one thing Steve Jobs / Apple is getting right. That
is getting out of the way and make *functionality* available to the average
user. And it's the *functionality* that sells these things, not the
technology. Technology is only the means to get the functionality to
the user. IMO this is the main thing engineers constantly fail to
understand. Users don't want technology, they want easy and intuitive
functionality available to them provided by whatever technology they may
end up with.

    But just because IPv6 has problems doesn't mean that it doesn't solve the fundamental address space problem in IPv4, and doesn't mean that it is anything less than the best available alternative.

What fundamental address space problem? I'd say we run out of AS numbers
about a year before we run out of IPv4 addresses, whenever that is.

What fundamental address space problem? I'd say we run out of AS numbers
about a year before we run out of IPv4 addresses, whenever that is.

Fortunately we have solutions for both. 32 bit ASNs and 128 bit
addresses. Pressure your vendors and peers to implement both.

Death of internet not predicted. No film at 11.

Rob

     IPv6 has its problems, yes. There are implementation issues that
confuse programmers at Sun working on Solaris, and confuse network
application programmers with a hell of a lot of experience under their
belt. If you can't talk directly to Jinmei himself, you're likely to be
well and truly screwed.

Ain't this *the* problem??? If not even Joe OperatingSystemEngineer
can understand it, what is John Doe at home supposed to do?

  John Doe at home is never going to see any of these issues. He'll see that BIND or NTP doesn't work correctly on his IPv6 implementation and then go to the appropriate mailing list or newsgroup and see that it works fine elsewhere, but that's as far as he'll go.

  Moreover, we're still in the very early phases of IPv6. We've learned how to move, but I'm not convinced that we're at the crawling stage, much less walking or running. There are a lot of issues that still have to be resolved.

  In comparison, where were we with IPv4 this many years after it was invented? We had, what, probably something less than 200 nodes on the ARPAnet, and DNS wasn't even a gleam in anyones eye?

  We're already way, way past that point with IPv6. Yes, we've got a long way to go, but we've also come a lot further a lot faster than anyone or anything else before. Give it a little time.

What fundamental address space problem? I'd say we run out of AS numbers
about a year before we run out of IPv4 addresses, whenever that is.

  AS numbers can be recycled. It is not politically feasible to insist that all those under-used address blocks get turned back in and more size-appropriate blocks get issued, so recycling of address blocks is both more difficult and happens more rarely.

  The problem with IPv4 space limitations is not the theoretical one of having more machines on the 'net than we can assign addresses to, although that problem will occur soon enough. The practical problem we have is that much of the address space has already been allocated, and was allocated in a manner that was not very space efficient, thus leaving us with a very nasty upcoming crunch.

  Now, if you honestly think you can solve that problem without going to an expanded address solution such as found in IPv6 (with IPv6 being the only practical model on the radar that I can see), then I would encourage you to do so and to report back when you're done.

I don't care what you see at RIPE meetings. You have to look at how
many servers/services are reachable via IPv6. Nothing else. Sure,
many European ISPs have got their IPv6 prefix and some even announce
it via BGP, but actually using it for anything more useful than "hey,
I can ping6 you!" is far off.

Well, a reasonable number of people are doing more than that. Of course I realize that the numbers that I'm about to list here can be interpreted in any number of ways, however, the trend is very clear: IPv6 is on the rise.

Once in a while when I have nothing to do I load up the Amsterdam Internet Exchange membership list and visit all the web sites from the members, while keeping an eye on tcpdump. This tells me how many of those member's web sites are reachable over IPv6. The latest numbers I have are for march 2005. At that time, 9 or 213 members had IPv6-enabled web sites. About a year earlier this was 4 or 5 (one had AAAA but was unreachable), no information on the then current number of members.

    IPv6 has its problems, yes. There are implementation issues that confuse programmers at Sun working on Solaris, and confuse network application programmers with a hell of a lot of experience under their belt. If you can't talk directly to Jinmei himself, you're likely to be well and truly screwed.

Ain't this *the* problem??? If not even Joe OperatingSystemEngineer
can understand it, what is John Doe at home supposed to do?

The trouble is that different OSes have different ideas about how you should deal with IPv4+IPv6 coexistance on the socket API level. This is a big headache for the unfortunate souls who have to deal with it, but it's of no consequence at all to users.

(If you write for one OS or one IP version or use a higher level API you won't have problems, though.)

You know, there is one thing Steve Jobs / Apple is getting right. That
is getting out of the way and make *functionality* available to the average
user.

I agree completely. All hail The Steve for giving us IPv6 on by default since MacOS 10.2! As of 10.4 the Safari browser handles IPv6 the way it should too, like iTunes and Apple's Mail have for ages (although there is a nasty bug in the 10.4 Mail that required me to go back to talking to my mail server over IPv4).

    But just because IPv6 has problems doesn't mean that it doesn't solve the fundamental address space problem in IPv4, and doesn't mean that it is anything less than the best available alternative.

What fundamental address space problem?

6 billion people with more under way with 3.7 billion usable addresses (how many do YOU use?) looks like a fundamental, long term problem to me.

I'd say we run out of AS numbers
about a year before we run out of IPv4 addresses, whenever that is.

The fix for this has been on the IETF drawing boards for half a decade but somehow seems to stay there.

OK, I'll bite.. :slight_smile:

Is there another alternative running around wearing a t-shirt that says
"I'm the best there is, but I'm not available"? :wink:

We are already behind in innovation as most networks these days are run by
accountants instead of people with an entrepaneur's sprit. We need good
business practices so that the network will stay afloat financially I do
not miss the 'dot.com' days.

But what we have now is an overemphasis on cost-cutting and like it or not
IPv6 implementation is seen as a 'frill' which will not reduce OPEX. I
really fear we have lost the edge here in the west due to too much
emphasis on the cost side of the equation ironically this has been driven
by the current network where financial information is available instantly
for decision making whereas in the past financial information about
far-flung operation took up to a year to to arrive so if a division was
profitable it was 'left alone' now with the instant availability we are
seeing profitable divisions of companies shut down because the numerical
analysis shows the capital could be used to generate a higher return
elsewhere.

Innovation is expensive and it does not return an immediate benefit and
right now all the average corporation cares about is the next quarter's
figures not whether the company will be profitable in 5 years. We are
seeing many instances of companies eating their seed corn instead of
investing in the future.

IPv6 would have been adopted much sooner if the protocol had been written
as an extension of IPv4 and in this case it could have slid in under the
accounting departments radar since new equipment and applications would
not be needed.

                            Scott C. McGrath

IPv6 would have been adopted much sooner if it had solved a problem that caused significant numbers of end users or large scale ISPs real pain. If IPv6 had actually addressed one or more of routing scalability, multi-homing, or transparent renumbering all the hand wringing about how the Asians and Europeans are going to overtake the US would not occur. Instead, IPv6 dealt with a problem that, for the most part, does not immediately affect the US market but which (arguably) does affect the other regions. I guess you can, if you like, blame it on the accountants...

Rgds,
-drc

You do make some good points as IPv6 does not address routing scalability
or multi-homing which would indeed make a contribution to lower OPEX and
be easier to 'sell' to the financial people.

As I read the spec it makes multi-homing more difficult since you are
expected to receive space only from your SP there will be no 'portable
assignments' as we know them today. If my reading of the spec is
incorrect someone please point me in the right direction.

IPv6's hex based nature is really a joy to work with IPv6 definitely fails
the human factors part of the equation.

                            Scott C. McGrath

Sliding anything past the accountants is bad practice. Is the goal to run IPv6 or to run a communications medium to support society? If it costs $1M to adopt IPv6 in the next quarter, what would you take the $1M from? (I used to work at a science research center. Having a good network wasn't the goal, doing science was. Without good science, there would be no FY++ budget for a better network.)

The Internet serves society, society owes nothing to the Internet. Members of this list may prioritize communications technology, other members of society may prioritize different interests and concerns. That is why IPv6 must offer a benefit greater than it's cost.

There is an element of fear-mongering in this discussion - that's why many
of us react poorly to the idea of IPv6. How so?

- We are running out of IPv4 space!
- We are falling behind <#insert scary group to reinforce fear of Other>!
- We are not on the technical cutting edge!

Fear is a convenient motivator when facts are lacking. I've read the above
three reasons, all of which are provable incorrect or simple fear mongering,
repeatedly. The assertions that we are falling behind the Chinese or
Japanese are weak echoes of past fears.

The market is our friend. Attempts to claim that technology trumps the
market end badly - anyone remember 2001? The market sees little value in v6
right now. The market likes NAT and multihoming, even if many of us don't.

Attempts to regulate IPv6 into use are as foolish as the use of fear-based
marketing. The gain is simply not worth the investment required.

- Daniel Golding

You are approaching the problem at the wrong end by asking "what's in it for me to adopt IPv6 now". The real question is "is IPv6 inevitable in the long run".

It's hard to be sure that the answer for that question is "yes", since all kinds of things can happen between now and, say, 2020. But it certainly looks like IPv4 addressing issues are becoming more and more painful over time. For instance, so far this year 98 million IPv4 addresses were assigned or allocated by RIRs. There are currently 1.1 - 1.2 billion usable addresses marked "reserved" (= "unused") by the IANA, so at this rate IANA be flat out in 2011. Now it's possible that the past 6 months were a fluke and it will take twice as long, or it's the start of a new trend and it's going to go even faster.

In any event, in the year 2020 we're NOT going to run IPv4 as we know it today. It's possible that the packets that travel over the wires still look like regular IPv4/TCP/UDP packets and all the complexity is pushed out to the application or political/economic layers, but that's not a possibility that appeals to me.

So by all means, be an IPv6 hold out as long as you like, but don't assume that just because adopting IPv6 doesn't make economic sense for you now, it isn't going to happen at some point in the next decade. No rush, though.

In any event, in the year 2020 we're NOT going to run IPv4 as we know
it today. It's possible that the packets that travel over the wires
still look like regular IPv4/TCP/UDP packets and all the complexity
is pushed out to the application or political/economic layers, but

                                      ^^^^^^^^^^^^^^^^^^^^^^^^^

that's not a possibility that appeals to me.

Is that layer 8?

Does anyone have a stateful firewall that works at that layer?

Cheers,
-- jra

With the chicken little-ing again...

You are approaching the problem at the wrong end by asking "what's in it for
me to adopt IPv6 now". The real question is "is IPv6 inevitable in the long
run".

Pardon my skepticism, but I recall hearing about the coming of the world due to pollution in the 1970's and the end of the oil supply by the 1980's. (E.g., see http://www.ncpa.org/pub/bg/bg159/ for a discussion on the latter, albeit written before the most recent oil 'scare.')

The point isn't whether IPv6 is good or not - it's that long-range predictions are often wrong. For every "memex" (http://www.iath.virginia.edu/elab/hfl0051.html) there's an oil crisis, Ada, GOSIP, economic default of New York City (Ford to City: Drop Dead! - NY Daily News, Oct 30, 1975)...

So by all means, be an IPv6 hold out as long as you like, but don't assume
that just because adopting IPv6 doesn't make economic sense for you now, it
isn't going to happen at some point in the next decade. No rush, though.

http://www.nanog.org/mtg-0405/augmentation.html

Been there, done that, documented and shared results. (Yes, got the T-Shirt too. It was a NANOG, after all.) That wasn't even the first go-round I had with IPv6.

My experiences were that IPv6 was painful - I ran into a lot of application bugs, OS's didn't deal with it well, and the ISP's were tough to deal with - as in, not many suppliers, not enough expertise to deliver on promises.

Maybe things are better now (note the use of past tense in the previous paragraph), I don't deal with IPv6 at this time.

With the chicken little-ing again...

?

You are approaching the problem at the wrong end by asking "what's in it for
me to adopt IPv6 now". The real question is "is IPv6 inevitable in the long
run".

Pardon my skepticism, but I recall hearing about the coming of the world due to pollution in the 1970's and the end of the oil supply by the 1980's.

That's nice, but maybe we should judge this issue own its own merits rather than adopt the position that since other people talking about other issues made mistakes in the past, surely there is a mistake this time too.

We know how many IPv4 addresses there are. We know how many are unusable (although this number isn't 100% fixed). We know how many were given out. We know how many are given out now each year. What kind of magic do you expect will make this problem that's coming go away?

And that's discounting that we already have a problem NOW. People are already moderating their requests because they know they can't get what they really want.

The point isn't whether IPv6 is good or not - it's that long-range predictions are often wrong.

It's very simple. IPv4 addresses will become scarce and expensive, unless either this internet fad blows over or a new technology replaces IPv4. Tell me how this "prediction" can be wrong. Are there hidden pockets of yet undiscovered address space? Is some government agency working on secret technology that lets you communicate over the net without the need for addresses?

My experiences were that IPv6 was painful - I ran into a lot of application bugs, OS's didn't deal with it well, and the ISP's were tough to deal with - as in, not many suppliers, not enough expertise to deliver on promises.

Maybe things are better now (note the use of past tense in the previous paragraph), I don't deal with IPv6 at this time.

It's getting better all the time, but there are still strange bugs in the applications, OSes and even the standards. IPv6 works very well for many things but not so well for others. Fortunately, there is still plenty of time to work out all the kinks before we need IPv6 to step up to the plate. In the mean time, we need SOME IPv6 so that the early adopters can find those kinks, and that part is right on track.

We who are running IPv6 salute you.

You do make some good points as IPv6 does not address routing scalability
or multi-homing which would indeed make a contribution to lower OPEX and
be easier to 'sell' to the financial people.

As I read the spec it makes multi-homing more difficult since you are
expected to receive space only from your SP there will be no 'portable
assignments' as we know them today. If my reading of the spec is
incorrect someone please point me in the right direction.

The spec in this case is RIR policy, which seems designed to accommodate the last-known word from the IETF on the subject, which is a pure aggregation model such as you described.

The fact that the pure aggregation model is insufficient in the real network has been widely recognised in IETF-land, and this was the reason that the multi6 working group was chartered. The multi6 working group produced a series of recommendations which in turn has led to the shim6 working group being formed. The shim6 working group has its first meeting in Paris in August.

If all this sounds like a lot of talking without much action then, well, yes. The problem being solved is not trivial, though, and shim6 is actually working towards something that could be implemented, rather than simply trying to throw ideas at the problem, so there is progress.

IPv6's hex based nature is really a joy to work with IPv6 definitely fails
the human factors part of the equation.

The phrase "IPv6's hex based nature" very pithily sums up the problem that IPv6 was designed to solve.

With great hindsight it would have been nice if the multi6/shim6 design exercise had come *during* the IPv6 design exercise, rather than afterwards: we might have ended up with a protocol/addressing model that accommodated both the address size problem and also the DFZ state bloat issue. Oh well.

Joe

Well, maybe I'm too optimistic here, but I believe that if a real solution to the DFZ problem presents itself, the IETF will bend over backwards and then some to shoehorn it into IP.

But it certainly looks like a small DFZ table and portable address space are fundamentally incompatible.

At least if you want all the advantages that real BGP multihoming has.
Not surprising. :slight_smile:

Best regards,
Daniel

Well, maybe I'm too optimistic here, but I believe that if a real solution to the DFZ problem presents itself, the IETF will bend over backwards and then some to shoehorn it into IP.

I'd say yes. You are too optimistic. :-).

But it certainly looks like a small DFZ table and portable address space are fundamentally incompatible.

Well, yes. Of course. If you make the routing locator also be the endpoint identifier, then _of course_ you must deal with the topological significance of the endpoint identifier. It sort of follows. You can't have your cake and eat it too.

Unfortunately, I do not believe a host-based solution like shim6 will ever be operationally deployable as it requires a rewrite of kernel stacks and such. I'm told people are already deploying IPv6 stacks that do not support the "mandatory" IPSEC goop and there is an expectation stack developers are going to tack on an optional bit of black magic that is used only in very rare circumstances? I have to admit some skepticism.

Rgds,
-drc