Discussion:
IPv6
(too old to reply)
Antonio Broughton
2003-07-16 05:04:47 UTC
Permalink
Hey, I was just looking through the archives of the list, and found a post
that was about IPv6 peering in New Zealand.

I am currently trying to setup a website (http://www.ipv6.co.nz), but as
such, I am the only one involved :P.

If anyone wants to help me setup content for the site etc, or if you would
like to use the domain for such things as the domain for a New Zealand IPv6
Peer... then by allmeans... you can use it, just email me...

Antonio Broughton
Joe Abley
2003-07-16 05:15:22 UTC
Permalink
Post by Antonio Broughton
Hey, I was just looking through the archives of the list, and found a post
that was about IPv6 peering in New Zealand.
I am currently trying to setup a website (http://www.ipv6.co.nz), but as
such, I am the only one involved :P.
If anyone wants to help me setup content for the site etc, or if you would
like to use the domain for such things as the domain for a New Zealand IPv6
Peer... then by allmeans... you can use it, just email me...
ISC is ready to peer over v6 at the APE, just as soon as there's a /64
there to number an interface in (and, preferably, at least one other
router attached to the APE to peer with :)

f.root-servers.net is reachable as 2001:500::1035.


Joe
Craig Whitmore
2003-07-16 05:30:26 UTC
Permalink
Post by Joe Abley
ISC is ready to peer over v6 at the APE, just as soon as there's a /64
there to number an interface in (and, preferably, at least one other
router attached to the APE to peer with :)
f.root-servers.net is reachable as 2001:500::1035.
Has Citylink got a IPv6 Peering Range for APE? WIX?

What other Providers are "thinking" about IPv6 (and multicast)

Thanks
Craig
Geoff Cant
2003-07-16 06:06:40 UTC
Permalink
Post by Craig Whitmore
Post by Joe Abley
ISC is ready to peer over v6 at the APE, just as soon as there's a /64
there to number an interface in (and, preferably, at least one other
router attached to the APE to peer with :)
f.root-servers.net is reachable as 2001:500::1035.
Has Citylink got a IPv6 Peering Range for APE? WIX?
What other Providers are "thinking" about IPv6 (and multicast)
Anyone know roughly how long this will take to filter down into ipv6
transport or tunnels for say people sitting on a jetstart connection?
--
Geoff Cant <***@uni.massey.ac.nz>
Simon Blake
2003-07-16 09:58:12 UTC
Permalink
Post by Craig Whitmore
Post by Joe Abley
ISC is ready to peer over v6 at the APE, just as soon as there's a /64
there to number an interface in (and, preferably, at least one other
router attached to the APE to peer with :)
f.root-servers.net is reachable as 2001:500::1035.
Has Citylink got a IPv6 Peering Range for APE? WIX?
Not right now, we don't, and unfortunately, unlike IPv4 we don't have
convenient unused allocations laying about looking for a good home on an
exchange, so I suspect we're going to have fess up with some cash.

If my reading of the APNIC site is correct, the smallest allocation we
can get will cost us US$1250. I have no problem with spending such a
sum - if a few ISP's say "yes, we're want to peer with IPV6, and we'll
pay some $ for the facility", then we'll go out and get an allocation
tomorrow.

Realistically, the $ are actually going to be pretty small - $50/month
from five ISP's would cover the APNIC cost, doubtless we'll find some
other reason to make it a more worthwhile sum :-).
Post by Craig Whitmore
What other Providers are "thinking" about IPv6 (and multicast)
When yawl progress beyond thinking and actually want to do it, sing out.
Despite what I've said above, it's likely that'll we'll have allocation
available for APE and WIX reasonably soon regardless. Serious requests
for peering will simply escalate the process.

Cheers
Si
Post by Craig Whitmore
Thanks
Craig
_______________________________________________
Nznog mailing list
http://list.waikato.ac.nz/mailman/listinfo/nznog
Nathan Ward
2003-07-16 21:47:08 UTC
Permalink
Post by Simon Blake
Post by Craig Whitmore
Post by Joe Abley
Post by Joe Abley
ISC is ready to peer over v6 at the APE, just as soon as there's a
/64
Post by Joe Abley
there to number an interface in (and, preferably, at least one other
router attached to the APE to peer with :)
Post by Joe Abley
f.root-servers.net is reachable as 2001:500::1035.
Has Citylink got a IPv6 Peering Range for APE? WIX?
Not right now, we don't, and unfortunately, unlike IPv4 we don't have
convenient unused allocations laying about looking for a good home on an
exchange, so I suspect we're going to have fess up with some cash.
If my reading of the APNIC site is correct, the smallest allocation we
can get will cost us US$1250. I have no problem with spending such a
sum - if a few ISP's say "yes, we're want to peer with IPV6, and we'll
pay some $ for the facility", then we'll go out and get an allocation
tomorrow.
<snip>
The following services are exempt from the IP resource application fee:
- IXP assignments
</snip>
--
Nathan Ward
Esphion Ltd.
Andy Linton
2003-07-18 01:31:10 UTC
Permalink
Post by Simon Blake
Post by Craig Whitmore
Post by Joe Abley
ISC is ready to peer over v6 at the APE, just as soon as there's a /64
there to number an interface in (and, preferably, at least one other
router attached to the APE to peer with :)
f.root-servers.net is reachable as 2001:500::1035.
Has Citylink got a IPv6 Peering Range for APE? WIX?
Not right now, we don't, and unfortunately, unlike IPv4 we don't have
convenient unused allocations laying about looking for a good home on an
exchange, so I suspect we're going to have fess up with some cash.
We've applied for IPv6 space for both APE and WIX. I'll report back here
once it's been approved.

andy

Steve Phillips
2003-07-16 11:05:56 UTC
Permalink
Post by Simon Blake
Not right now, we don't, and unfortunately, unlike IPv4 we don't have
convenient unused allocations laying about looking for a good home on an
exchange, so I suspect we're going to have fess up with some cash.
If my reading of the APNIC site is correct, the smallest allocation we
can get will cost us US$1250. I have no problem with spending such a
sum - if a few ISP's say "yes, we're want to peer with IPV6, and we'll
pay some $ for the facility", then we'll go out and get an allocation
tomorrow.
I know of at least three ISP's that are actively looking into IPv6 but the
main issue they face is router upgrades to support it and currently the
"looking" involves things such as "wtf ?? thats an IP address ??" (along
with another two following the 'but no one wants it' scenario)

Today I was looking at setting up a 6to4 relay router and the steps
involved but no one seems to know anyone that can help from an ISP
perspective and no one so far has been able to give any pointers toward
this end effect (its only been a day tho so there is still hope ! :-) ).
Post by Simon Blake
Realistically, the $ are actually going to be pretty small - $50/month
from five ISP's would cover the APNIC cost, doubtless we'll find some
other reason to make it a more worthwhile sum :-).
Had not looked at the costings so far but apparently you need to prove that
there is enough interest such that you would hand out at least 200 /48's
worth of space within the next two years - It may pay to hold back until
the equipment is enabled and possibly practice with some equiv. RFC1918
type space first and then look toward a neutral party applying and acting
as an LIR initially.

Billing may end up being a rather big part of this as well and there are a
lot of "eek !" type things when one starts looking into this from an IP
billing perspective, and this could end up being a big part of things were
6to4 gateways provided for much of New Zealand.
Post by Simon Blake
Post by Craig Whitmore
What other Providers are "thinking" about IPv6 (and multicast)
When yawl progress beyond thinking and actually want to do it, sing out.
Despite what I've said above, it's likely that'll we'll have allocation
available for APE and WIX reasonably soon regardless. Serious requests
for peering will simply escalate the process.
I'll add you to my list :-) only been looking for two weeks now and already
have 3 ISP's that are making serious noises about it.

I think part of the issue is that people see it as a rather almighty big
step to take in learning new things [tm] and this puts a lot of people
off.. I went from knowing nothing over a weekend to having established a
tunnel to my mates place and having two subnets of three machines each
talking to each other, doing DNS, web, ssh and e-mail quite happily and
with a mixture of Win2k, Linux and WinXP boxes and I'm not even a routing
guy, so it cant be that hard ! (took around a day in total)

If more people start looking into the requirements at least to ensure their
core network will support it (IOS versions, switch and router upgrades etc)
then we may be in a better position to gauge what is involved in getting
some rudimentary peering working.
--
Steve.

PS: Anyone got any jobs going ? :-)
Barry Murphy
2003-07-16 05:23:58 UTC
Permalink
Hey Antonia,

Just on that note, I will be adding ipv6 to the site sometime soon, when i'm
not busy. I have the IP addresses, but was wondering if anyone would be
interested in peering.

Barry

----- Original Message -----
From: "Antonio Broughton" <***@world-net.co.nz>
To: <***@list.waikato.ac.nz>
Sent: Wednesday, July 16, 2003 5:04 PM
Subject: [nznog] IPv6
Post by Antonio Broughton
Hey, I was just looking through the archives of the list, and found a post
that was about IPv6 peering in New Zealand.
I am currently trying to setup a website (http://www.ipv6.co.nz), but as
such, I am the only one involved :P.
If anyone wants to help me setup content for the site etc, or if you would
like to use the domain for such things as the domain for a New Zealand IPv6
Peer... then by allmeans... you can use it, just email me...
Antonio Broughton
_______________________________________________
Nznog mailing list
http://list.waikato.ac.nz/mailman/listinfo/nznog
Nathan Ward
2003-07-16 06:00:31 UTC
Permalink
Post by Barry Murphy
Hey Antonia,
Just on that note, I will be adding ipv6 to the site sometime soon, when i'm
not busy. I have the IP addresses, but was wondering if anyone would be
interested in peering.
Barry
<rant>
Insert generic chicken egg story here. (1)

If we all start tunnelling, there may be a bit more of the learning taking
place that would be needed for said ISPs to roll out v6.
Perhaps WLUG's metanet(2) could be utilised here? Thier v6 support is
progressing it seems.
However, the metanet requires a proprietary piece of software that AFAIK is
better optimised for a emulating a single ethernet broadcast domain. (ie
peer to peer, stuff like that) ie. it doesn't talk to your Cisco's, or your
Junipers(3).

However. with a number of geeks running on metanet, perhaps machines could
be put at APE and WIX to translate between metanet and native. There are a
few technical issues to sort out here but most of them are trivial.
Perhaps these machines could also terminate GRE tunnels or <insert v6 over
v4 tunneling protocol here that works on many vendor's routers>, so that
companies/people that can't either get there natively, or don't have linux
routers things, can.

This kind of scenario would allow many of the geeks around to get onto v6,
and get learning it in a 'read world' (4) environment whilst ISPs and so on
talk about doing it. When they get around to it, there will already be a
(small) userbase.

And because Craig brought it up, multicast. Multicast gateways from APE/WIX
onto the metanet. Sure, it makes it being multicast somewhat pointless (4),
but again, its to build a userbase, and get some 'real world' (4) learning
done.
--
Nathan Ward

(1) We've heard it all before. 'no address space problem' 'no userbase'
(2) I'm not part of metanet or WLUG, so take this all as 42'nd hand
information
(3) WAND will run on FreeBSD AFAIK...
(4) Chaotic, changing, non-controlled environment.
(5) Actually, a little less, as there are some providers who have 'native'
multicast to thier customers.
James Spooner
2003-07-16 06:12:28 UTC
Permalink
Well since Nathans gone and told everyone :)
Post by Nathan Ward
If we all start tunnelling, there may be a bit more of the learning taking
place that would be needed for said ISPs to roll out v6.
Perhaps WLUG's metanet(2) could be utilised here? Thier v6 support is
progressing it seems.
http://www.wlug.org.nz/MetaNet
http://www.wlug.org.nz/MetaNetIPv6
Post by Nathan Ward
However, the metanet requires a proprietary piece of software that AFAIK is
better optimised for a emulating a single ethernet broadcast domain. (ie
peer to peer, stuff like that) ie. it doesn't talk to your Cisco's, or your
Junipers(3).
We have a daemon called Etud that simulates a broadcast ethernet using UDP.
This basically means any machine can 'establish' a tunnel with another,
without
the need for a single tunnel endpoint like FreeNet.

We run BGP over this ethernet so it's a virtual peering point, using unicast
v4 to send
packets between the two metanet boxes, which means v6 performance is
(almost) as good
as native v4.
Post by Nathan Ward
However. with a number of geeks running on metanet, perhaps machines could
be put at APE and WIX to translate between metanet and native. There are a
few technical issues to sort out here but most of them are trivial.
Perhaps these machines could also terminate GRE tunnels or <insert v6 over
v4 tunneling protocol here that works on many vendor's routers>, so that
companies/people that can't either get there natively, or don't have linux
routers things, can.
We already have a GRE? tunnel from a cisco box to a well-connected box on
metanet.
Post by Nathan Ward
This kind of scenario would allow many of the geeks around to get onto v6,
and get learning it in a 'read world' (4) environment whilst ISPs and so on
talk about doing it. When they get around to it, there will already be a
(small) userbase.
The best thing about it is that we don't have to endure the insane latency
FreeNet6 brings.

James Spooner
Joe Abley
2003-07-16 06:09:08 UTC
Permalink
Post by Geoff Cant
Post by Craig Whitmore
Post by Joe Abley
ISC is ready to peer over v6 at the APE, just as soon as there's a
/64
there to number an interface in (and, preferably, at least one other
router attached to the APE to peer with :)
f.root-servers.net is reachable as 2001:500::1035.
Has Citylink got a IPv6 Peering Range for APE? WIX?
What other Providers are "thinking" about IPv6 (and multicast)
Anyone know roughly how long this will take to filter down into ipv6
transport or tunnels for say people sitting on a jetstart connection?
You can't plumb tunnels right now, today? Or is the problem that
there's just no local place to plumb a tunnel to?

(see http://ipv6tb.he.net/ for a tunnel broker run by clueful people in
the US).


Joe
Barry Murphy
2003-07-16 06:14:09 UTC
Permalink
I have the tunnel, but I dont see the purpose in going international to
connect to a NZ server. If I was to use the ISC f root server with ipv6, my
ping times would be higher than if I were to use the root server in Asia
because it would be routing back to NZ after the international hops.

B.m


----- Original Message -----
From: "Joe Abley" <***@isc.org>
To: "Geoff Cant" <z-***@inspire.net.nz>
Cc: <***@list.waikato.ac.nz>
Sent: Wednesday, July 16, 2003 6:09 PM
Subject: Re: [nznog] IPv6
Post by Joe Abley
Post by Geoff Cant
Post by Craig Whitmore
Post by Joe Abley
ISC is ready to peer over v6 at the APE, just as soon as there's a
/64
there to number an interface in (and, preferably, at least one other
router attached to the APE to peer with :)
f.root-servers.net is reachable as 2001:500::1035.
Has Citylink got a IPv6 Peering Range for APE? WIX?
What other Providers are "thinking" about IPv6 (and multicast)
Anyone know roughly how long this will take to filter down into ipv6
transport or tunnels for say people sitting on a jetstart connection?
You can't plumb tunnels right now, today? Or is the problem that
there's just no local place to plumb a tunnel to?
(see http://ipv6tb.he.net/ for a tunnel broker run by clueful people in
the US).
Joe
_______________________________________________
Nznog mailing list
http://list.waikato.ac.nz/mailman/listinfo/nznog
Geoff Cant
2003-07-16 06:14:56 UTC
Permalink
<snip>
Post by Joe Abley
Post by Geoff Cant
Anyone know roughly how long this will take to filter down into ipv6
transport or tunnels for say people sitting on a jetstart connection?
You can't plumb tunnels right now, today? Or is the problem that
there's just no local place to plumb a tunnel to?
(see http://ipv6tb.he.net/ for a tunnel broker run by clueful people
in the US).
Er, I can probably tunnel to freenet6 and so on, but I guess I'm more
interested in an NZ provider (the better to natter to them when things
go wrong :)
--
Geoff Cant <***@uni.massey.ac.nz>
p***@deeper.co.nz
2003-07-16 09:42:48 UTC
Permalink
Post by Joe Abley
You can't plumb tunnels right now, today? Or is the problem that
there's just no local place to plumb a tunnel to?
(see http://ipv6tb.he.net/ for a tunnel broker run by clueful people in
the US).
ahh, but tunnels to the US mean that you have the irritating problem
that pinging someone on the same ISP as you can give you up to 3,000ms
of latency as it transits from here to the US and back again. You have
issues when they are overloaded, or down. You have issues if the
internet between here and there decides that packets should be routed
via darkest peru. IPv6 was designed to work in the situation where you
have to bridge large IPv4 clouds.

A much better solution for people in New Zealand is to use 6to4
addressing, especially if you're on the end of ADSL or similar and don't
have native transport.

6to4 addressing means that whatever device(s) have non-RFC1918 IPv4
addresses can have a /48 assigned to them. 2003:<high two octets of
IPv4 address>:<low two octets of IPv4 address>::/48. To route to
another 2002::/16 address over IPv4 you just take the packet,
encapsulate it in IPv4, and send it to the embedded IPv4 address. Linux
and FreeBSD both support this. Latency to other 6to4 addresses is about
the same as IPv4, avoiding the entire US round trip.

To talk from a 6to4 address to a non-6to4 IPv6 address you route via
::192.88.99.1 which is an AnyCast address. Unfortunately, last time I
looked the nearest 192.88.99.1 AnyCast announcement appeared to be in
Germany. Is there any chance of someone putting forward 6to4 anycast
gateway in the APE and/or the WIX? Of course the problem here is that
probably 50% of it's traffic will be international.

In an unrelated note, the MetaNet has been used by several people in NZ
who are interested in playing around with more interesting networks
without breaking things that people pay to use. We use it for peering
IPv6 without having to go via the US and we'd be interested in organising
IPv6 tunnels with like minded people (either using wand, gre, or 6in4
tunnels) Unfortunately the MetaNet wouldn't handle multicast very well
(or we'd be having a good tinker with it)

Resources:
* http://www.wlug.org.nz/DancingPenguin (A IPv6 enabled website in New
Zealand which is useful for testing with)
* http://www.wlug.org.nz/IPv6LessonsLearnt (For some random things we've
discovered about IPv6 the hard way, so you don't have to)
* http://www.wlug.org.nz/6to4 (Information about 6to4 under Linux)
* http://www.wlug.org.nz/IPv6 (Information about IPv6 on the wiki)
* http://www.nevada.net.nz/~pmurray/6to4.html (6to4 under FreeBSD)

If you want to chat about it, we're usually hanging around in #wlug on
Undernet. irc://undernet/%23wlug

- --
<f00Dave> Look, rejects, this is #OpenGL, not #GEEKSEX.
Perry Lorier
2003-07-16 08:30:11 UTC
Permalink
Post by Joe Abley
You can't plumb tunnels right now, today? Or is the problem that
there's just no local place to plumb a tunnel to?
(see http://ipv6tb.he.net/ for a tunnel broker run by clueful people in
the US).
ahh, but tunnels to the US mean that you have the irritating problem
that pinging someone on the same ISP as you can give you up to 3,000ms
of latency as it transits from here to the US and back again. You have
issues when they are overloaded, or down. You have issues if the
internet between here and there decides that packets should be routed
via darkest peru. IPv6 was designed to work in the situation where you
have to bridge large IPv4 clouds.

A much better solution for people in New Zealand is to use 6to4
addressing, especially if you're on the end of ADSL or similar and don't
have native transport.

6to4 addressing means that whatever device(s) have non-RFC1918 IPv4
addresses can have a /48 assigned to them. 2003:<high two octets of
IPv4 address>:<low two octets of IPv4 address>::/48. To route to
another 2002::/16 address over IPv4 you just take the packet,
encapsulate it in IPv4, and send it to the embedded IPv4 address. Linux
and FreeBSD both support this. Latency to other 6to4 addresses is about
the same as IPv4, avoiding the entire US round trip.

To talk from a 6to4 address to a non-6to4 IPv6 address you route via
::192.88.99.1 which is an AnyCast address. Unfortunately, last time I
looked the nearest 192.88.99.1 AnyCast announcement appeared to be in
Germany. Is there any chance of someone putting forward 6to4 anycast
gateway in the APE and/or the WIX? Of course the problem here is that
probably 50% of it's traffic will be international.

In an unrelated note, the MetaNet has been used by several people in NZ
who are interested in playing around with more interesting networks
without breaking things that people pay to use. We use it for peering
IPv6 without having to go via the US and we'd be interested in organising
IPv6 tunnels with like minded people (either using wand, gre, or 6in4
tunnels) Unfortunately the MetaNet wouldn't handle multicast very well
(or we'd be having a good tinker with it)

Resources:
* http://www.wlug.org.nz/DancingPenguin (A IPv6 enabled website in New
Zealand which is useful for testing with)
* http://www.wlug.org.nz/IPv6LessonsLearnt (For some random things we've
discovered about IPv6 the hard way, so you don't have to)
* http://www.wlug.org.nz/6to4 (Information about 6to4 under Linux)
* http://www.wlug.org.nz/IPv6 (Information about IPv6 on the wiki)
* http://www.nevada.net.nz/~pmurray/6to4.html (6to4 under FreeBSD)

If you want to chat about it, we're usually hanging around in #wlug on
Undernet. irc://undernet/%23wlug

- --
<f00Dave> Look, rejects, this is #OpenGL, not #GEEKSEX.
Joe Abley
2003-07-16 06:15:56 UTC
Permalink
Post by Barry Murphy
I have the tunnel, but I dont see the purpose in going international to
connect to a NZ server. If I was to use the ISC f root server with
ipv6, my
ping times would be higher than if I were to use the root server in
Asia
because it would be routing back to NZ after the international hops.
OK. We have a strong preference not to use tunnels to talk v6 to
people, but we will happily peer with anybody who can talk to us
directly over the APE.

Perhaps this is the point where someone offers to run a
MetaNet-connected box attached to the APE, or some other APE-connected
tunnel broker.


Joe
Barry Murphy
2003-07-16 06:52:11 UTC
Permalink
I have ipv6 through freenet and I have a box housed at orcon who peer to
ape. Whats involved with the metanet installation on a freebsd box?

Barry

----- Original Message -----
From: "Joe Abley" <***@isc.org>
To: "Barry Murphy" <***@unix.co.nz>
Cc: "Geoff Cant" <z-***@inspire.net.nz>; <***@list.waikato.ac.nz>
Sent: Wednesday, July 16, 2003 6:15 PM
Subject: Re: [nznog] IPv6
Post by Joe Abley
Post by Barry Murphy
I have the tunnel, but I dont see the purpose in going international to
connect to a NZ server. If I was to use the ISC f root server with
ipv6, my
ping times would be higher than if I were to use the root server in
Asia
because it would be routing back to NZ after the international hops.
OK. We have a strong preference not to use tunnels to talk v6 to
people, but we will happily peer with anybody who can talk to us
directly over the APE.
Perhaps this is the point where someone offers to run a
MetaNet-connected box attached to the APE, or some other APE-connected
tunnel broker.
Joe
James Spooner
2003-07-16 06:59:58 UTC
Permalink
----- Original Message -----
From: "Barry Murphy" <***@unix.co.nz>
To: "Joe Abley" <***@isc.org>
Cc: <***@list.waikato.ac.nz>
Sent: Wednesday, July 16, 2003 6:52 PM
Subject: Re: [nznog] IPv6
Post by Barry Murphy
I have ipv6 through freenet and I have a box housed at orcon who peer to
ape. Whats involved with the metanet installation on a freebsd box?
Not too much, we have a version of Etud that can be compiled on
freebsd 4.8?, I can organise it for you if you like.
Joe Abley
2003-07-16 07:04:23 UTC
Permalink
Post by Barry Murphy
I have ipv6 through freenet and I have a box housed at orcon who peer
to
ape. Whats involved with the metanet installation on a freebsd box?
Note that the orcon router sitting on the APE will need v6 transport,
and routing back to the freenet box, if you want to be able to reach F
locally over v6.
Nathan Ward
2003-07-16 22:26:56 UTC
Permalink
Post by Barry Murphy
I have ipv6 through freenet and I have a box housed at orcon who peer to
ape. Whats involved with the metanet installation on a freebsd box?
Note that the orcon router sitting on the APE will need v6 transport, and
routing back to the freenet box, if you want to be able to reach F
locally over v6.
Alternatively, I'm /sure/ the community can donate some hardware to put a
box /at/ APE. So we aren't reliant on any ISP. And we don't cause any
unneeded traffic for them either.

Infact I've got a box right here. And I'd even be willing to find some cash
to get hold of one of those rack cases from Roger.

Of course, such a box would probably have to have a 'geek project'
disclaimer on it so that people don't do anything remotely critical. And
all kinds of things like that.

--
Nathan Ward
Esphion Ltd.
Roger De Salis
2003-07-16 23:46:25 UTC
Permalink
Post by Nathan Ward
Alternatively, I'm /sure/ the community can donate some hardware to put
a box /at/ APE. So we aren't reliant on any ISP. And we don't cause any
unneeded traffic for them either.
Infact I've got a box right here. And I'd even be willing to find some
cash to get hold of one of those rack cases from Roger.
I will donate cases to this worthy cause, in the same manner
as the root servers. As everyone else
contributes along the way, we can get this done....

R
Post by Nathan Ward
Of course, such a box would probably have to have a 'geek project'
disclaimer on it so that people don't do anything remotely critical. And
all kinds of things like that.
Nathan Ward
Esphion Ltd.
--
\_ Roger De Salis ***@fx.net.nz
</' FX Networks +64 25 481 452
/) Wellington, New Zealand ***@desalis.gen.nz
(/
` Bentley win Le Mans! - What next, English win cricket in Aus?
Steve Phillips
2003-07-17 00:42:09 UTC
Permalink
Post by Nathan Ward
Post by Barry Murphy
I have ipv6 through freenet and I have a box housed at orcon who peer to
ape. Whats involved with the metanet installation on a freebsd box?
Note that the orcon router sitting on the APE will need v6 transport, and
routing back to the freenet box, if you want to be able to reach F
locally over v6.
Alternatively, I'm /sure/ the community can donate some hardware to put a
box /at/ APE. So we aren't reliant on any ISP. And we don't cause any
unneeded traffic for them either.
Part of the idea would be to build a native IPv6 network in New Zealand, as
such - the ISP's will need to be reasonably heavily involved as they will
have issues they will need to sort out.

I am a little confused how "having a box sitting at APE" will actually
assist toward anything with respect to IPv6 networks in New Zealand.

In order to access this box (not using tunnels) one would assume that the
link terminating equipment at APE for [insert company name here] would
support IPv6 and be able to do peering with another peer on APE, hence
negating the requirement for a 'box sitting at ape' all together..

I would assume we could get this done initially with private v6 ranges and
then look toward getting a publicly routable range. However, in order to
link back into the rest of the world someone would be required to put up
bandwidth to do this and as such, it would be unfair to expect them to
carry all the costs for everyone (hence a system to apportion cost, i.e - a
billing system)

I think it is hard enough to just get people to support the idea of IPv6
let alone starting to talk about donating equipment and the like. I think
InternetNZ was also looking at putting together a komittee of some form to
investigate IPv6 in New Zealand and assist in rolling it out ? possibly
this would be a good place to start and get interested (and in some cases
not so interested) parties involved.

Can anyone comment on where things are with respect to this komittee that
was mentioned a while back on NZNOG and how one could express an interest
in sitting in on it ?
--
Steve.
Nathan Ward
2003-07-17 01:05:12 UTC
Permalink
Post by Steve Phillips
Part of the idea would be to build a native IPv6 network in New Zealand,
as such - the ISP's will need to be reasonably heavily involved as they
will have issues they will need to sort out.
I have a native v6 network. The issue is connecting that to other native v6
networks to get some usage out of it. ISPs sit around talking about doing
native v6 end to end.
I am proposing that while we wait for that to happen on a large scale, we
go and setup a peering/tunnel point for those of us at the end of v4 only
networks.
Post by Steve Phillips
I am a little confused how "having a box sitting at APE" will actually
assist toward anything with respect to IPv6 networks in New Zealand.
It will get people using and learning v6 on a larger scale, and having the
box at APE will allow ISPs to turn on native v6 and still talk to everyone
already using this tunnelled network.
Post by Steve Phillips
In order to access this box (not using tunnels) one would assume that the
link terminating equipment at APE for [insert company name here] would
support IPv6 and be able to do peering with another peer on APE, hence
negating the requirement for a 'box sitting at ape' all together..
Even if said box doesn't peer with anyone to start with, its still
progress. It allows us to use v6 now, and it provides a very simple
migration path to native v6 for APE connect ISPs. For non APE connected
companies, ISPs etc, they can 'upgrade' now without having to wait for
thier upstream providers first.
Post by Steve Phillips
I would assume we could get this done initially with private v6 ranges
and then look toward getting a publicly routable range. However, in order
to link back into the rest of the world someone would be required to put
up bandwidth to do this and as such, it would be unfair to expect them to
carry all the costs for everyone (hence a system to apportion cost, i.e -
a billing system)
This is a peering point to improve national v6 speed/RTTs.
Users of the system provide thier own address space from (insert tunnel
broker here) or APNIC.

Such a box would only provide free fast national v6 connectivity. Put it at
APE so that NZ ISPs currently talking about rolling out v6 can have thier
shiny new v6 networks accessable via this v6 IX.
--
Nathan Ward
Esphion Ltd.
James Spooner
2003-07-17 01:14:07 UTC
Permalink
Post by Steve Phillips
I would assume we could get this done initially with private v6 ranges and
then look toward getting a publicly routable range. However, in order to
link back into the rest of the world someone would be required to put up
bandwidth to do this and as such, it would be unfair to expect them to
carry all the costs for everyone (hence a system to apportion cost, i.e - a
billing system)
A point to consider here is that there _is no such thing_ as a 'private'
IPV6
range in my understanding. When we began peering over MetaNet we used
Freenet as an address broker, whether or not we intended on keeping our
links to
freenet active, as it was the only (somewhat) permenant way of getting
addresses
that other people arn't using.

James
Steve Phillips
2003-07-17 01:30:47 UTC
Permalink
Post by Nathan Ward
Post by Steve Phillips
Part of the idea would be to build a native IPv6 network in New Zealand,
as such - the ISP's will need to be reasonably heavily involved as they
will have issues they will need to sort out.
I have a native v6 network. The issue is connecting that to other native
v6 networks to get some usage out of it. ISPs sit around talking about
doing native v6 end to end.
I am proposing that while we wait for that to happen on a large scale, we
go and setup a peering/tunnel point for those of us at the end of v4 only
networks.
It would be better to pressure your ISP into doing something about IPv6
even if this is simply "please setup a tunnel end point for people to use !"

Putting the tunnel endpoint on APE would potentially cause issues (routing
could be fun if it was not to have access internationally - and even the
"citylink local to NZ" ranges cause problems for the large majority of NZ)

As previously stated, I have been looking as the "second step" to putting a
local tunnel endpoint in New Zealand somewhere that will be open to people
wanting to peer with it via tunnels, this would simply be a normal box
housed at some shonky ISP that didn't mind forking out small amounts for
international bandwidth and was still connected reasonably well in New
Zealand nationally so as not to worry about national capacity.

This still does not address the main issue, and that is - ISP's are not
moving to IPv6 because /insert latest reason not to here/

Getting native peering would be a good first step toward acceptance and
judging community support, end to end connectivity options may be the next
and then putting pressure on the likes of Telecom and Telstra to accept
routes across their backbones would be the last i assume. but being only a
lowly Systems Engineer I am not really in the position to make informed
decisions about stuff like that and it may be better to get a komittee to
look over it and suggest a road plan going forward.
--
Steve.
Nathan Ward
2003-07-17 02:23:55 UTC
Permalink
Post by Steve Phillips
Post by Nathan Ward
Post by Steve Phillips
Part of the idea would be to build a native IPv6 network in New
Zealand, as such - the ISP's will need to be reasonably heavily
involved as they will have issues they will need to sort out.
I have a native v6 network. The issue is connecting that to other native
v6 networks to get some usage out of it. ISPs sit around talking about
doing native v6 end to end.
I am proposing that while we wait for that to happen on a large scale,
we go and setup a peering/tunnel point for those of us at the end of v4
only networks.
It would be better to pressure your ISP into doing something about IPv6
even if this is simply "please setup a tunnel end point for people to use !"
How is that better? That doesn't allow people from different ISPs to
communicate unless there is inter-ISP tunneling going on as well. And
again, it relies on the ISP doing something. I am thinking ISP independant,
they can't charge for it, they can't turn it off, they can't neglect it
because it doesn't make them money.
Post by Steve Phillips
Putting the tunnel endpoint on APE would potentially cause issues
(routing could be fun if it was not to have access internationally - and
even the "citylink local to NZ" ranges cause problems for the large
majority of NZ)
I don't follow.. How will that cause issues? (I'm not aware of this
'citylink local to nz' range issue you speak of)
Post by Steve Phillips
As previously stated, I have been looking as the "second step" to putting
a local tunnel endpoint in New Zealand somewhere that will be open to
people wanting to peer with it via tunnels, this would simply be a normal
box housed at some shonky ISP that didn't mind forking out small amounts
for international bandwidth and was still connected reasonably well in
New Zealand nationally so as not to worry about national capacity.
Why does there have to be this international thing?
Sure, a .nz tunnel provider with international connectivity would be nice,
but is it really useful if we have a 'local addresses only' tunnel
endpoint? There are already international tunnel brokers around..
Any who do you suggest? I don't see any hands being raised. And if lots of
people start using it, is it going to stay there?
Post by Steve Phillips
This still does not address the main issue, and that is - ISP's are not
moving to IPv6 because /insert latest reason not to here/
It may address some of these issues. For instance - Users, there will now
be people using v6 in NZ. Waiting for upstream providers, there are many
non-APE connected ISPs. They can now offer v6 ( and are more likely to than
larger providers, because upgrading a smaller network is alot easier and
cheaper ).

Who cares? If we want to do it, lets just do it. As I said, its a geek
project for the moment, but when the time eventually comes that the ISPs
get over all those reasons, there will be people using v6 in New Zealand
already who will want to sign up for connectivity/services with native v6.
--
Nathan Ward
Esphion Ltd.
Steve Phillips
2003-07-17 01:36:20 UTC
Permalink
Post by James Spooner
Post by Steve Phillips
I would assume we could get this done initially with private v6 ranges and
then look toward getting a publicly routable range. However, in order to
link back into the rest of the world someone would be required to put up
bandwidth to do this and as such, it would be unfair to expect them to
carry all the costs for everyone (hence a system to apportion cost, i.e -
a
Post by Steve Phillips
billing system)
A point to consider here is that there _is no such thing_ as a 'private'
IPV6
range in my understanding. When we began peering over MetaNet we used
Freenet as an address broker, whether or not we intended on keeping our
links to
freenet active, as it was the only (somewhat) permenant way of getting
addresses
that other people arn't using.
the fe{c,d,e,f}x are reserved as site-local ranges and (according to the
FAQ) were likened to the RFC1918 address space.

[quote]

3.2.2. Site local address type

These are addresses similar to the
<http://www.faqs.org/rfcs/rfc1918.html>RFC 1918 / Address Allocation for
Private Internets in IPv4 today, with the added advantage that everyone who
use this address type has the capability to use the given 16 bits for a
maximum number of 65536 subnets. Comparable with the 10.0.0.0/8 in IPv4 today.

Another advantage: because it's possible to assign more than one address to
an interface with IPv6, you can also assign such a site local address in
addition to a global one.

It begins with:

fecx: <- most commonly used.
fedx:
feex:
fefx:


(where "x" is any hex character, normally "0")

[/quote]

taken from: http://www.tldp.org/HOWTO/Linux+IPv6-HOWTO/x479.html

as such, I'd suggest possibly playing with these for a start and then
working upward once people are familiar with the fact that a 128bit IP
address is pretty much the same as a 32 bit one with more bits..
--
Steve.
DPF
2003-07-17 08:51:19 UTC
Permalink
Post by Steve Phillips
I think it is hard enough to just get people to support the idea of IPv6
let alone starting to talk about donating equipment and the like. I think
InternetNZ was also looking at putting together a komittee of some form to
investigate IPv6 in New Zealand and assist in rolling it out ? possibly
this would be a good place to start and get interested (and in some cases
not so interested) parties involved.
Can anyone comment on where things are with respect to this komittee that
was mentioned a while back on NZNOG and how one could express an interest
in sitting in on it ?
INZ has just had its AGM and in the next week will be finalising its
committee memberships, including that of Technical Committee which
this falls under.

The plan as I understand it is for Technical Committee to setup and
support a number of Taskforces (or Tiger Teams as Keith calls them) in
areas such as IPv6, Enum, Broadband, DNSSEC etc.

Offers to help out will be warmly welcomed I am sure. I suspect a
general call for interested persons will happen sometime in August -
but that's up to the Chair of Technical.

DPF
--
E-mail: ***@farrar.com
ICQ: 29964527
MSN: ***@hotmail.com
Joe Abley
2003-07-16 09:48:40 UTC
Permalink
Post by p***@deeper.co.nz
To talk from a 6to4 address to a non-6to4 IPv6 address you route via
::192.88.99.1 which is an AnyCast address. Unfortunately, last time I
looked the nearest 192.88.99.1 AnyCast announcement appeared to be in
Germany. Is there any chance of someone putting forward 6to4 anycast
gateway in the APE and/or the WIX? Of course the problem here is that
probably 50% of it's traffic will be international.
... in which case it makes more sense for individual ISPs to install
their own 6to4 gateways for the use of their customers, and not
announce it to peers or transit providers, so that the international
traffic it sinks or generates is attributable to a paying customer.

This still relies on some v6 infrastructure to be deployed in the ISP,
and the general domestic traffic issue (for all v6 destinations)
remains unless there's some domestic v6 peering.


Joe
Chris Hellberg
2003-07-16 09:49:56 UTC
Permalink
when I was running the v6ix in Wellington I was running EBGP sessions with most people who had connections; seemed to work well when I had a bunch of open-souce software mirrors which were v6-accessible and people had routes that went straight to the IX.

There was also tunnel to the 6bone at Merit, for those who wanted to have a nosey at the 6bone routing table, although there were some 2000::/24 prefixes on the 6bone.

I've talked to Si at Citylink about getting a PI peering block for WIX and APE, but it wasn't a huge priority. I used addresses from a freenet allocation and carved up a /64 in to a bunch of /126s and used them as point-to-point links. Not the best solution in the world, but worked for the purpose.

-----Original Message-----
From: ***@deeper.co.nz [mailto:***@deeper.co.nz]
Sent: Wednesday, 16 July 2003 21:43
To: Joe Abley
Cc: ***@list.waikato.ac.nz
Subject: Re: [nznog] IPv6


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Post by Joe Abley
You can't plumb tunnels right now, today? Or is the problem that
there's just no local place to plumb a tunnel to?
(see http://ipv6tb.he.net/ for a tunnel broker run by clueful people in
the US).
ahh, but tunnels to the US mean that you have the irritating problem
that pinging someone on the same ISP as you can give you up to 3,000ms
of latency as it transits from here to the US and back again. You have
issues when they are overloaded, or down. You have issues if the
internet between here and there decides that packets should be routed
via darkest peru. IPv6 was designed to work in the situation where you
have to bridge large IPv4 clouds.

A much better solution for people in New Zealand is to use 6to4
addressing, especially if you're on the end of ADSL or similar and don't
have native transport.

6to4 addressing means that whatever device(s) have non-RFC1918 IPv4
addresses can have a /48 assigned to them. 2003:<high two octets of
IPv4 address>:<low two octets of IPv4 address>::/48. To route to
another 2002::/16 address over IPv4 you just take the packet,
encapsulate it in IPv4, and send it to the embedded IPv4 address. Linux
and FreeBSD both support this. Latency to other 6to4 addresses is about
the same as IPv4, avoiding the entire US round trip.

To talk from a 6to4 address to a non-6to4 IPv6 address you route via
::192.88.99.1 which is an AnyCast address. Unfortunately, last time I
looked the nearest 192.88.99.1 AnyCast announcement appeared to be in
Germany. Is there any chance of someone putting forward 6to4 anycast
gateway in the APE and/or the WIX? Of course the problem here is that
probably 50% of it's traffic will be international.

In an unrelated note, the MetaNet has been used by several people in NZ
who are interested in playing around with more interesting networks
without breaking things that people pay to use. We use it for peering
IPv6 without having to go via the US and we'd be interested in organising
IPv6 tunnels with like minded people (either using wand, gre, or 6in4
tunnels) Unfortunately the MetaNet wouldn't handle multicast very well
(or we'd be having a good tinker with it)

Resources:
* http://www.wlug.org.nz/DancingPenguin (A IPv6 enabled website in New
Zealand which is useful for testing with)
* http://www.wlug.org.nz/IPv6LessonsLearnt (For some random things we've
discovered about IPv6 the hard way, so you don't have to)
* http://www.wlug.org.nz/6to4 (Information about 6to4 under Linux)
* http://www.wlug.org.nz/IPv6 (Information about IPv6 on the wiki)
* http://www.nevada.net.nz/~pmurray/6to4.html (6to4 under FreeBSD)

If you want to chat about it, we're usually hanging around in #wlug on
Undernet. irc://undernet/%23wlug

- --
<f00Dave> Look, rejects, this is #OpenGL, not #GEEKSEX.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)
Comment: Only when you are sure they have you, can you stop being paranoid

iD8DBQE/FQ0TcAgRpy8z8UQRAhadAJ4gAGn+YRABqt6A7X58K73hBQPNugCgy3ie
05qO9xRxAXyaG1K5EElVfAY=
=+lkH
-----END PGP SIGNATURE-----
_______________________________________________
Nznog mailing list
***@list.waikato.ac.nz
http://list.waikato.ac.nz/mailman/listinfo/nznog
Joe Abley
2003-07-17 01:30:39 UTC
Permalink
On Wednesday, Jul 16, 2003, at 21:14 Canada/Eastern, James Spooner
Post by James Spooner
A point to consider here is that there _is no such thing_ as a
'private'
IPV6
range in my understanding.
There are link-local addresses, which are somewhat similar in concept.
There was a proposal to specify a site-local address format too, but
that got shot down at the v6ops meeting in San Francisco earlier this
year.

If MetaNet provides a single-subnet, multi-access shared medium
abstraction, then a link local addressing scheme might be good for
something. Bill Manning once popularised a scheme for using link-local
addressing at exchange points in order to avoid RIR policy, to the
point of running code even, but it was not widely adopted for reasons
that escape me.


Joe
James Spooner
2003-07-17 01:45:23 UTC
Permalink
Post by Joe Abley
If MetaNet provides a single-subnet, multi-access shared medium
abstraction, then a link local addressing scheme might be good for
something. Bill Manning once popularised a scheme for using link-local
addressing at exchange points in order to avoid RIR policy, to the
point of running code even, but it was not widely adopted for reasons
that escape me.
We tried using link locals as addresses to peer with each other. The
problem with link locals is that you have to instruct all your software as
to which interface a particular link local address is on. I'm not sure how
well BGP's next-hop calculations work in this case, at one stage we were
getting the next-hop being set to the peer always. Generally link locals
seem to be more trouble than they're worth for setting up permenant routing
infrastructures.

Cheers

James
Joe Abley
2003-07-17 02:05:04 UTC
Permalink
On Wednesday, Jul 16, 2003, at 21:36 Canada/Eastern, Steve Phillips
Post by Steve Phillips
the fe{c,d,e,f}x are reserved as site-local ranges and (according to
the FAQ) were likened to the RFC1918 address space.
http://www.psg.com/~mrw/ipv6-wg-minutes-mar2003.txt

(look for "Site-Local Usage Discussion", then the discussion that
Post by Steve Phillips
Margaret asks the question, do we want to deprecate site local
addresses?
There are 20 hands up for not deprecating, 102 hands up for
deprecating;
This is interpreted as rough consensus to deprecate site local
addressing; this consensus will have to be verified on the mailing
list.
)


Joe
Continue reading on narkive:
Loading...