[6bone] report of 6bone planning BOF

[Jeroen Massar]

Jonathan Guthrie [mailto:jguthrie@brokersys.com] wrote:

> On Wed, Mar 19, 2003 at 11:10:33PM +0100, Jeroen Massar wrote:
> > Jonathan Guthrie wrote:
>  
> > > On Wed, Mar 19, 2003 at 10:35:08AM -0800, Bob Fink wrote:
>   
> > > > The 6bone phaseout proposal was discussed with a strong 
> > > consensus reached 
> > > > on a pTLA allocation cutoff date of Jan 1, 2004 and a 6bone 
> > > turn off date 
> > > > of June 6, 2006 (6/6/6).
>  
> > > I think it's insane to set a date for the end of the 6bone 
> > > when there is no replacement for it in the USA.
>  
> > Remind me again about why americans are allowed to only
> > think of theirselves.
> 
> Why American's are allowed to only think of themselves?  Are you
> telling me that you don't use ANY services that have any presence at
> all in the US?  That's a pretty tall claim.  This Internet thing is
> about communications.  Communications implies that both ends can,
> well communicate.  Cutting off one side or another arbitrarily will
> result in reduced value to both sides.

Fortunatly the Internet is global and doesn't entirely
rely on the US. And it isn't 'cutting' off, current 6bone
users have 3 long years the time to make sure they have
RIR space connectivity. Seeing the history of computing,
a lot can happen in 3 years.

> For what it's worth, it seems to me that in the whole "let's schedule
> the destruction of the 6bone" discussion, the commentary has been
> dominated by people in Europe how have been thinking only of 
> themselves.

Never heard of Asia now did you? That is the place where
IPv6 is *it*. You might note that 6bone chairs and other
coordinators are all US based. Not european. Also like you
say yourself, there have been a lot of voices from the
european 'side' because these people do voice themselves.
Apparently the american 'side' doesn't have an interest
in the 6bone or silently agree with it and want to move forward.

You should realize that the 6bone is a *TEST* network
for testing out new things. It's not for production networks,
like providing paying clients with IPv6. ISP's tend to be
production, at least I hope so for their customers.
You are implying that you want a production connection,
well then you won't need the 6bone for that now do you?

> > There are many countries which
> > don't have IPv6 at all.
> 
> Then the date for destroying the nest of tunnels that is the 6bone
> should not be scheduled until all of them have native IPv6 service,
> too.  I didn't write about them because I can't speak for them, I
> can only speak for myself.

6bone == testing, you are talking about production space.
Production space can be arranged in all countries on this planet
through the RIR's (ARIN, APNIC & RIPE).
Though LACNIC doesn't have a clear policy set out yet.
But they are working on it.

> > But the US certainly has 
> > *Grab stats*
> 
> > 76
> 
> > Now say again that there is no IPv6 in the US?
> 
> Okay.  There are no IPv6 providers in the US.  That I can find.  I've
> been talking about this for a month, now, and nobody on this list has
> attempted to enlighten me so I conclude that nobody else knows of any
> in the US, either.  You certainly don't, or you would have named one.
> I've looked, and I haven't been able to find any.

You are looking for enduser IPv6 native connectivity.
Now everybody can tell you that, except for asia, this is mostly
not possible. Why? well for the exact reasons that you mention: 
No hardware support to the enduser. That's why there are transition
methods available.

> Note, please, that the fact that somebody has chosen to 
> purchase a block of addresses does not imply that they
> have connectivity for sale or even that they intend to.

Ofcourse, and even announcing it into the DFZ doesn't mean
that they are using it either, let alone that somebody is
watching over it. I am quite aware of that.

> What I really need is 2B ISDN dial-up with fixed addresses.  However,
> even if someone is selling this service it doesn't matter because I
> know of no equipment I can use for my end on that sort of connection.
> I will admit that I haven't looked too hard for that.  (What would be
> the point?)

Move to japan, they have it over there, but then again you won't
find much ISDN over there as it's mostly cable (16mbit for 60$ US).
Now wonder again why they win at the world CyberGames ;)

> > Maybe you should start kicking around your ISP's a bit more?
> 
> "Kicking around"?  Just exactly WHAT kind of pull am I expected to
> have with my ISP?  As I've said before, threatening to leave won't
> cut it.  "Give me IPv6 connectivity or I'll disconnect completely from
> the Internet" isn't a credible threat.

I quote from http://ipv6.he.net
8<--------------------------------------
Hurricane Electric's support for IPv6
Hurricane Electric is currently running a production IPv6 network and
offering business class commercial IPv6 services. Native IPv6
connectivity is available for both direct connection customers and
colocation customers. Hurricane Electric also provides a free tunnel
broker which allows users to experiment with IPv6 by tunneling over the
existing IPv4 Internet. Hurricane Electric's tunnel broker is available
for use by anybody.
-------------------------------------->8

And I am european and not even related to he.net.
In a way they are 'competition' for me, mind you,
though I don't think of it that way. Note that HE has both
RIR and 6bone space and so have many others. 

> > There are enough ways of getting IPv6 in europe so why
> > should that be so different from the US?
> 
> I don't know.  My problem is that nobody who is pushing a rapid
> destruction of the 6bone realizes that the conditions are different
> in places that aren't right exactly where they are.
> 
> If I were to be forced to guess, I'd guess that the real problem is
> the amount of money invested in current infrastructure that would
> have to be written off unless the equipment purchased by that money
> can be somehow upgraded.  From what I can see, the Lucent Max TNT
> that I connect to cannot.

That is exactly what the 'problem' is. Money. But why should that
inhibit a pending closure, which is only going to happen
in 3 years of a network (6bone) which is meant for testing?

> > You should also note that 2006 is in 3 years. There
> > will change a load of things before that date.
> 
> There will?  Maybe, maybe not.  Your crystal ball isn't any 
> clearer than mine.  I predict that the ILEC-based ISPs and
> the cable companies will have the bulk of the US Internet
> access business in 2006 and none of those companies has
> shown the slightest interest in IPv6 or has any
> intention of moving any faster than they possibly can on this issue.

Then, if you are so determined to get *production* IPv6
at home, you should be talking to them why they are not
doing it, when you got them in the boat it will all go faster.

Also if we're taking it from that side I see a huge
business opportunity in setting up a IPv6 provider in the US. <hint> :)

> Of course, when you really get right down to it, what I really object
> to is setting a calendar date on something that doesn't really have a
> time element to it.  What is necessary to make the dismantling of the
> 6bone a good idea is about how widely IPv6 is deployed in production,
> not about when some date rolls around.   That's why setting a date,
> even one far in the future, until the basic conditions for making the
> 6bone irrelevant is not well considered.

Apparently there are not many americans who complain about this.
Also I think that most part of the _endusers_, including me and
a load of others all around this globe, are still using tunneled
connectivity. There are a happy view who have native connectivity
ofcourse but I don't think it will be expressable in percentages.

Why do you think we build a whitelabeled tunnelbroker system
which provides RIR and 6bone on some POPs space to endusers.
The target is to have a POP at an ISP so that it's clients
can get IPv6 connectivity from their own ISP. Yes, tunneled.
But without upgrading the intermediate hardware, which is
usually quite a burden to replace, financially mostly.

Let's take a looky at my home network:
http://purgatory.unfix.org/network.gif

See that nice thing called "Alcatel SpeedTouch", well it
bridges my ethernet into the Cistron router. So purgatory
simply has the MAC of the cistron router (195.62.92.1) in
it's table. Effectively if that router supported IPv6 it
could simply give me IPv6 by doing some ND and RA.
Unfortunatly that box doesn't do IPv6, neither does the box
that is connecting it to the rest of world which is behind
it. The trick is that, quite close (~2ms max) is another
router which plays POP and which gives me near-native IPv6:
For example compare these two:

jeroen@purgatory:~$ traceroute6 newszilla6.xs4all.nl
traceroute to newszilla6.xs4all.nl (2001:888:0:4::119) from
3ffe:8114:2000:240:290:27ff:fe24:c19f, 30 hops max, 16 byte packets
 1  gw-20.ams-02.nl.sixxs.net (3ffe:8114:1000::26)  20.02 ms  32.728 ms
19.626 ms
 2  Amsterdam.core.ipv6.intouch.net (2001:6e0::2)  19.549 ms  19.775 ms
20.078 ms
 3  ams-ix.tc2.xs4all.net (2001:7f8:1::a500:3265:2)  20.774 ms  19.838
ms  36.207 ms
 4  0.ge-0-1-0.xr1.pbw.xs4all.net (2001:888:0:103::1)  22.904 ms  35.862
ms  20.324 ms
 5  newszilla6.xs4all.nl (2001:888:0:4::119)  20.162 ms  20.637 ms
20.164 ms

jeroen@purgatory:~$ traceroute newszilla.xs4all.nl
traceroute to newszilla.xs4all.nl (194.109.133.20), 64 hops max, 40 byte
packets
 1  gw-64-92.sms-1.ams-tel.cistron.net (195.64.92.1)  18 ms  17 ms  18
ms
 2  ve10.rtr-1.ams-tel.cistron.net (62.216.31.1)  18 ms  18 ms  18 ms
 3  ams-ix.tc2.xs4all.net (193.148.15.166)  19 ms  18 ms  18 ms
 4  0.ge-0-1-0.xr1.pbw.xs4all.net (194.109.5.1)  24 ms  18 ms  18 ms
 5  194.109.133.2 (194.109.133.2)  19 ms  19 ms  19 ms

IPv4 path almost matches IPv6 path.
As you see I also use 6bone space and that will end on 06/06/2006.
I want it to end it before that too and get production space
with support and all other benefits. Playtime is over.
Though I do have to note that my IPv6 tunnel has been up for
over about 2 years now so it's stable enough to be called
production.

By 2006 I hope the above mentioned boxes have been replaced
by correct hardware which can handle IPv6. Otherwise I
guess that my upstream ISP has installed their own POP
and is nicely handing out 2001:9b8::/32 space to their clients.
Then again who says that I am still using their service,
or that my ISP is still in business, that they upgraded to ADSL2 etc?
Or that we are simply internetting over WLAN? New technologies
are bound to get IPv6 supported hardware.

By the way there is something which is much worse about all this.
Why doesn't for example Google, Altavista, CNN etc don't have IPv6 yet?
Ai, but that just might roll into the chicken and egg problem again.

I hope that clears up my viewpoint a bit.
In short: 6bone = testing, it's now time for production.

Greets,
 Jeroen