Whirlpool

Hi all,

Thought you'd like to know first – some press will be issued later today.

---

Today Internode is making IPv6 services available to its national customer base.

Internode is now running a native IPv6 network spanning our national network in Australia and its international PoPs (actually, we have been for a while).

According to the sixxs.net web site (http://www.sixxs.net/faq/connectivity/?faq=ipv6transit), we're the only (and hence by definition, the first) Australian transit provider to offer commercial IPv6 transit to customers on a national basis in Australia.

(On a non-commercial basis, AARNet have had IPv6 running for some time, and there is at least one international IPv6 transit provider in Sydney, but we believe we're the first national ISP to run a full IPv6 backbone and to open it up to all of our customers).

To do this, the Internode core routers are running in 'dual stack' mode, offering both IPv4 and IPv6 services, and are natively linked to IPv6 transit providers we use in our international PoP sites.

As with IPv4, we'll be happy to peer with other like minded IPv6 national backbones (as soon as there are any!).

As of today most customers connected via ethernet or fibre, or co-located, can receive a supported native IPv6 connection if they require one.

We have had some trial customers running in this mode for some months now – it all seems to work :)

ADSL and other broadband customers nationally customers that wish to start exploring IPv6 can do so now, connecting via an Internode tunnel broker service, with 'best effort' support available for those customers. That service is running on a commercial IPv6 tunnel broker appliance from Hexago, which is in turn plugged into our native IPv6 backbone (i.e. the tunnel ends at the Hexago and from then on its all native IPv6).

FYI, we are also working toward offering direct native IPv6 access over layer 2 (i.e. directly to an ADSL router or similar) in the future. That isn't available generally today (though we do have it working in the lab).

In parallel we're continuing efforts to convince more ADSL router vendors to support native IPv6 in their products... today its really just Cisco routers and some custom builds of Linux kernel replacements for one or two consumer routers that can support native layer 2 IPv6 access at all.

Meantime, again, the tunnel broker is the way to get connected to IPv6 without your ADSL router needing to support it (yet).

For more information, see our new customer IPv6 home page:

http://ipv6.internode.on.net/

(and please do read all of the information on that page before diving in; If you're not comfortable loading up and configuring the tunnel client software, then don't do so – in this regard, IPv6 is still only just emerging from 'early adopter' status, and if you want to wait for more general deployment, then wait for the native layer 2 access and for native layer 2 support in your router).

Regards,
Simon Hackett

Does the sentence (under Unmetering and IPv6) :

"Internode has not yet determined whether unmetering will be either technically feasible or commercially viable to offer to its customers in the future."

only apply to IPv6, or also to IPv4?

I don't know a great deal about this, but does this mean that people running IPv6 sit in a separate bandwidth pool for international traffic as you may not be able to access all the overseas providers you might otherwise have access to using IPv4?

nigel writes...

only apply to IPv6

IPv6 – nothing has changed IPv4 wise.

nigel writes...

only apply to IPv6, or also to IPv4?

v6

EDIT – bah beaten :-P

Mick writes...

but does this mean that people running IPv6 sit in a separate bandwidth pool for international traffic as you may not be able to access all the overseas providers you might otherwise have access to using IPv4?

Same network, different packets – no different "pool" as our core network is IPv6 from Perth to Tokyo and LA/San Jose.

Only difference is that the number of people using IPv6 is less. But you'll be dual stack, so hey – give it a go.

Mick writes...

I don't know a great deal about this, but does this mean that people running IPv6 sit in a separate bandwidth pool for international traffic as you may not be able to access all the overseas providers you might otherwise have access to using IPv4?

Its all the same big happy network. They're all packets. There is no issue about a 'bandwidth pool' here. Its the same physical links.

Obviously not every upstream provider in the USA and Japan and not every peering point in Australia is yet IPv6 enabled. But if its there – we'll peer with it if and when its IPv6 capable.

You may also not appreciate what 'dual stack' means – IPv6 is a completely new Internet to IPv4 in many respects. Parallel and distinct. So (for instance) data paths won't necessarily be the same to reach a given city/country/whatever – now or in the future. Both the IPv4 and IPv6 internets will exist for a long time in parallel.

Simon

Very cool news guys. I've been using my Apple Airport to give my home network IPv6 access via a 6to4 tunnel. ATM this means my v6 traffic is tromboning through Switzerland.

Anyone know if the airport can be used to connect to a hexago tunnel broker? I'll have to give it a go when I get home.

Simon Hackett writes...

http://ipv6.internode.on.net/

Would love to see the HTTP logs for this site. It would be interesting to track the percentage of visitors from v4/6to4/internode v6/non-internode v6 addresses over time.

yob writes...

Anyone know if the airport can be used to connect to a hexago tunnel broker? I'll have to give it a go when I get home.

Not sure, but please do give it a go – if it works, I'll do the same thing at my place :)

Simon

Cool dunno if I need it but I think I will check it out anyway.
:)

When ipv6 is supported on adsl routers will they be assigned dynamic or static ipv6 addresses by default? ie, home plan gets static not just soho/business.

With the number of available ipv6 addresses I cant see why you would need dynamic addresses.

Foggy2 writes...

When ipv6 is supported on adsl routers will they be assigned dynamic or static ipv6 addresses by default?

Haven't had the opportunity to work that out yet, because there isn't any consumer-grade IPv6-capable CPE yet so we haven't needed to apply any brainspace to dealing with policies associated with connecting it. :-)

On a semi-related point: Don't get too attached to the subnets the Hexago tunnel broker allocates to you. They won't generally change, but we reserve the right to blow 'em away and reallocate them if required.

– mark

Took me a while to get it working.
Currently running Win XP SP 3 (will configure linux too).
I found that I needed to download Gateway6 6.0-BETA4 to get it to run with XP SP3.
now see what I can do with this thing. :)

Oh PS thanks Internode as well.
:)

Two questions:

1) I'm currently on iinet but I have access to an internode ADSL account. Can I connect to the tunnel broker with this username/password on my iinet service?
2) If 1) works, does the IPv6 traffic count on the internode account?

mwdmeyer writes...

Can I connect to the tunnel broker with this username/password on my iinet service?

You're not meant to be able to. It's intended to be on-net access only.

Nice one guys... some movement on this is great, hopefully others will follow (again.. teehee.. gg 'node)

Congratulations on your achievement. :)

mwdmeyer writes...

1) I'm currently on iinet but I have access to an internode ADSL account. Can I connect to the tunnel broker with this username/password on my iinet service?

You just need to apply a patch to that iiNet broadband service first, by clicking here:

https://secure.internode.on.net/webtools/ids-home

It will take just a few days for the patch to be applied and then you'll have access from that location to IPv6 :)

Matthew Moyle-Croft writes...

You're not meant to be able to. It's intended to be on-net access only.

You need to tweak the http://ipv6.internode.on.net/ page a bit then. Currently there's a heading that says Who can get it? which then explains how to get it.

Can I just ask, what difference will it make to the regular customer? The only thing that I can see applying to me is all content being metered, which is bad.

You guys rock ( the Australian broadband scene)!

Knight of Cydonia writes...

I can see applying to me is all content being metered, which is bad.

You can easily revert back to ipv4 (disconnect ipv6) via the gateway 6 client program before you go into unmetered content. Although just have to remember to switch back to ipv4 or else your unmetered downloads will be counted.

What gives me the poops is that Sol still won over Simon on TeleAwards on CEO of the year.

What a con!. :/

Ted Bullpitt writes...

Although just have to remember to switch back to ipv4 or else your unmetered downloads will be counted.

It's worth noting that mirror.internode.on.net is the IPv4 address and mirror.ipv6.internode.on.net is the IPv6 address – so it's pretty clear about what's what.

It's also worth REPEATING – it's not a comment on the FUTURE plans or not of us with regards to metering, merely that we've not explored it yet and whether it can/can't be done etc.

Knight of Cydonia writes...

Can I just ask, what difference will it make to the regular customer? The only thing that I can see applying to me is all content being metered, which is bad.

In 3-4 years time when we run out of IPv4 addresses, your inter-tubes will keep working. Other than that, there's no practical advantage to end users.

Right now, there's really *very* little motivation to use Internode's v6 offering at home other than teaching yourself how it works. The few pieces of major content available over v6 (like Internode's mirror, google search and the Bejing olympics site) are all still available via v4 addresses.

I think the best thing about this move is that it shows Internode are preparing their network, staff and tools for the transition. The better they understand v6, the less chance end users will have to worry about the upheaval over the next few years.

Edit: expanded my opinion a little

yob writes...

In 3-4 years time when we run out of IPv4 addresses, your inter-tubes will keep working. Other than that, there's no practical advantage to end users.

Except that 'node will have ironed out many more issues/bugs than competitors. When IPv4 exhaustion hits it will be ugly for about 12-18 months.

Consumer equipment suppliers will be loving it.

Knight of Cydonia writes...

an I just ask, what difference will it make to the regular customer?

Access to the new, different, parallel Internet running on a new protocol layer.

If that doesn't appeal, thats fine – don't do it (yet) :)

The only thing that I can see applying to me is all content being metered, which is bad.

Not correct.

If you access IPv4 content via IPv4 then unmetering (where offered) continues to apply- even if you have an IPv6 tunnel turned on, and (obviously) providing you access the resources using the IPv4 native path.

But if you choose to turn on IPv6, you are the one that needs to use appropriate care in that regard, not us.

All we do is meter and measure the IPv4 packets.

And all tunnelled IPv6 is accounted for as metered IPv4 tunnel packets, downloaded by you along with all your other normal access.

In the future, when we have dual stack layer 2 access available, we'll be in a position to reconsider the situation regarding unmetering again.

Meantime, if this bothers you – then don't do it. That simple.

Simon

Great news.
and congrats on this job well done

Well done Internode :)

DISREGARD FIXED IT
ADMIN FEEL FREE TO DELEYE THIS POST

however help needed

C:\>ipv6 install
Installing...
Failed to complete the action.
Error 0x800704b8
C:\>netsh int ipv6 install
An extended error has occurred.
C:\>

the-real-flip writes...

however help needed

http://store.apple.com/au

the-real-flip writes...

however help needed

C:\>ipv6 install
Installing...
Failed to complete the action.
Error 0x800704b8

Googling the error message indicates that this URL might help you http://www.sixxs.net/forum/?msg=setup-228387

Daryl

Anyone know the support if this work on a Linksys Wrt54g with many custom firmware distros?

yob writes...

Would love to see the HTTP logs for this site. It would be interesting to track the percentage of visitors from v4/6to4/internode v6/non-internode v6 addresses over time.

We watching this carefully too. We may release the statistics in the future.

Thought I check it out so I installed the client and when I try to connect it says tunnel creation script failed to execute, so I looked at the logs and it doesnt say much? Would the router firewall be blocking it?

Edit:Typos

Congratulations! You're viewing this page using IPv6 ( 2001:44b8:61::2b ) !!!!!

yay :)

i think for some people, the non-free content will be the big decision in (not) changing over.

my 2c

Eevo writes...

i think for some people, the non-free content will be the big decision in (not) changing over.

Can you guys PLEASE just get over this bit? If you run the tunnel client and you access IPv4 content it's still free.

What are the benefits from us ipv6?

will it make my pages load faster?

Just curious!

Will give it a go when im not on the train on the way home lol.

Thanks node.

Eevo writes...

i think for some people, the non-free content will be the big decision in (not) changing over.

Its not either/or. You can run both concurrently. And you can switch the IPv6 tunnel on and off on demand.

If a given source is only running IPv4, then you'll get the data via IPv4. Likewise for a pure IPv6 host; There is only an ambiguity to be careful of when an unmetered data source is fielded under the same name for both protocols.

This is why we have actually got IPv6 running on the mirror server, for instance, but only on a distinct hostname, to let you control the outcome in that regard by choosing which name you connect to (and as a result, if you access it as 'mirror.internode.on.net' then its coming in via IPv4 and hence remains unmetered).

You have nothing but improved choices here – not increased constraints.

But its fine – noone is forcing you (yet; There will be a day when no more IPv4 addresses are available and then people will be 'suddenly' very interested indeed – and at that point, well, Internode customers – and staff – will be well ahead of the game)

staogswolbloS writes...

What are the benefits from us ipv6?

Right now? Being a part of the future before your mates.

will it make my pages load faster?

No.

And... as an added bonus, it is completely free of any added Yak Fat.

Take a read of the IPv6 information pages here, to get some idea of what this is all about:

http://ipv6.internode.on.net/sites/

Simon Hackett writes...

But its fine – noone is forcing you (yet; There will be a day when no more IPv4 addresses are available and then people will be 'suddenly' very interested indeed – and at that point, well, Internode customers – and staff – will be well ahead of the game)

So basically it is just another ip protocol that will allow us to have another ip address when we run out of them?

No real enhancement to our connection then?

Oh well thanks again node at least I know I will always have an ip and once again node is the leader of the pack.

Ted Bullpitt writes...

Currently running Win XP SP 3 (will configure linux too).

I found that I needed to download Gateway6 6.0-BETA4 to get it to run with XP SP3.

I just installed this and it seems to be working...just, but keep "Aquiring IP address" even as I am typing this.. btw it drops then starts again looking for IP address.
Any ideas I can try?

The only thing that concerns me is this:

No Unmetering with IPv6

The 'Unmetering' of certain sites that is offered to Internode ADSL customers is not available via IPv6 (regardless of access method). Internode has not yet determined whether unmetering will be either technically feasible or commercially viable to offer to its customers in the future.

Accordingly, all downloads of data from the Internet via IPv6 is metered, regardless of whether a given site is available on an 'unmetered' basis to ADSL customers using IPv4.

Hopefully some content stays unmetered, for the day in the future when a node DSLAM arrives at my exchange.... *hint hint*

? Matt ? writes...

Hopefully some content stays unmetered

Well currently if your still accessing it via IPv4, it will still be metered..

This is why they have seperated the DNS entries for their mirror, one's just mirror.internode.on.net the other is mirror.ipv6.internode.on.net

http://ipv6.internode.on.net/configuration/tunnel-windows/

Download the file from http://go6.net/4105/download.asp, and select the Gateway6 Client 5.1 Windows Installer 32-bit.

Its a small point, but obviously get the 64-bit installer if you have 64 bit Vista.

Will this help at all with NAT issues from having multiple machines connected to one ADSL router ?

? Matt ? writes...

The only thing that concerns me is this:

Please read Simon and my comments in this thread.

fabricator writes...

Will this help at all with NAT issues from having multiple machines connected to one ADSL router ?

Possibly, but probably not. Depends on how the consumer router manufacturers decide to setup their devices.

By default, I suspect most will have an SPI firewall, so no traffic can get in unless an internal machine initiated the connection (just like NAT!), or a permanent rule is setup to allow traffic through.

So even though all machines will have globally routable IPv6 addresses, holes would still need to be punched in the firewall for things like VoIP and BT.

Disclaimer: all my opinion, maybe I'll be proved wrong.

Well Done!!! Leading the way yet again! :)

Thanks Simon!

Great stuff, Internode is on top as it has always been!
However, I am not sure how this will impact the end-user who typically uses a router which does not support IPv6.

Regards
Cliff

"It's likely that IPv6 will give us enough addresses for quite some time – about 340,000,000,000,000,000,000,000,000,000,000,000,000 in fact."

What's that, 340 gajillion squillion?

Simon Hackett writes...

Right now? Being a part of the future before your mates.

Too right. This is seriously cool. I have used the aarnet broker for a while, and the thought of the IPV6 tunnel getting shorter (to the extent that it disappears altogether) is very cool.

I hope the Billion folks are busy cutting some new firmware code to get their routers ready for a native IPv6 tail. One willing beta tester over here.

And to the people continually banging on about free this, or unmetered content that, give it a break. Its as bad as out and out trolling. Just go back under your rock and stop knocking those that are want to drive change (and those who want to jump on-board and give the "change bus" a push).

In the meantime im preying that the future goodness of IPv6 from Internode will also apply to those of us on a Telstra RIM!

This is great.

There's a few mistakes in the Mac directions at http://ipv6.internode.on.net/configuration/tunnel-macos/

Type cd ~/Desktop/gw6c-5_1-RELEASE-src and press enter to change into the source folder.

In Leopard, the default download location is ~/Downloads/, not Desktop.

Type the command chmod +x /usr/local/g26c/template/*.sh. This will alter the permission of any file name with a .sh extension.

g26c should be gw6c.

Works though:

x-2:~ gm$ ping6 www.ipv6.org
PING6(56=40+8+8 bytes) 2001:44b8:61::55 --> 2001:6b0:1:ea:202:a5ff:fecd:13a6
16 bytes from 2001:6b0:1:ea:202:a5ff:fecd:13a6, icmp_seq=0 hlim=45 time=413.574 ms
16 bytes from 2001:6b0:1:ea:202:a5ff:fecd:13a6, icmp_seq=1 hlim=45 time=408.307 ms
16 bytes from 2001:6b0:1:ea:202:a5ff:fecd:13a6, icmp_seq=2 hlim=45 time=414.253 ms
16 bytes from 2001:6b0:1:ea:202:a5ff:fecd:13a6, icmp_seq=3 hlim=45 time=408.672 ms
^C
--- shake.stacken.kth.se ping6 statistics ---
4 packets transmitted, 4 packets received, 0% packet loss
round-trip min/avg/max = 408.307/411.202/414.253 ms

:D

cliff101 writes...

However, I am not sure how this will impact the end-user who typically uses a router which does not support IPv6.

Just go and re-read the parts about the Tunnel Broker :)

t3h writes...

In Leopard, the default download location is ~/Downloads/, not Desktop.

Yep.

Also, at least in the gw6 beta, the config file is installed with perms 700 (if you're on a multi-user system and don't want others knowing your Internode password, this makes sense), so you need to be root to edit it and run the client as root, or change the permissions.

I also found out that it's quite important to follow the instruction on disabling ipv6 on all interfaces, otherwise the tun/tap kernel module dies (woops).

16 bytes from 2001:44b8:8020:f501:250:56ff:feb3:6633, icmp_seq=0 hlim=60 time=72.363 ms

:)

wow it works nicely.

Shame i would never be able to remember my ip it's so long.

Thanks Simon.

Hi,

I disconnected it but for some reason ipconfig says I have 3 network connections and ip is still ipv6.

I have uninstalled the program.

staogswolbloS writes...

I disconnected it but for some reason ipconfig says I have 3 network connections and ip is still ipv6.

I have uninstalled the program.

What happens when you reboot?

Same thing.

Now says ipv4 ip address is 10.0.0.1. lol

Though still says I have netork connection 7 with a ipv6 ip.

I have disabled ipv6 on my network card and rebooted and get the same thing.

EDIT: Seems ok I can't view ipv6 websites.

staogswolbloS writes...

I disconnected it but for some reason ipconfig says I have 3 network connections and ip is still ipv6.

When you install the Hexago software in Windows it enables IPv6 in Windows. This will stay enabled until you disable it. That interface with an IPv6 address is likely your LAN adapter with an auto-configured address, because IPv6 is now enabled.

staogswolbloS writes...

I disconnected it but for some reason ipconfig says I have 3 network connections

It works by creating an extra "tunnel" connection.

and ip is still ipv6.

Are you sure? Try a shift-reload on the ipv6.internode.on.net site to see if it drops back. I suspect the program may leave the tunnel interface there even though it breaks the connection. You can also force windows to drop the IPv6 ip address:
ipconfig /release interfacename.

Michael Kratz writes...

That interface with an IPv6 address is likely your LAN adapter with an auto-configured address, because IPv6 is now enabled.

I would have thought it would just list the IPv6 IP address on the same interface, and that the new one is the tunnel interface, but then it's been a whilr since I mucked around with IP settings on Windows.

Matt writes...

Are you sure? Try a shift-reload on the ipv6.internode.on.net site to see if it drops back. I suspect the program may leave the tunnel interface there even though it breaks the connection. You can also force windows to drop the IPv6 ip address:

ipconfig /release interfacename.

Yeah I just did that but still says my ip is 10.0.0.1.

I think that ipv6 address it shows in ipconfig is for reference when it is turned on or something.

yob writes...

By default, I suspect most will have an SPI firewall, so no traffic can get in unless an internal machine initiated the connection (just like NAT!), or a permanent rule is setup to allow traffic through.

Does anyone know where to look for ideas on how to tighten up a Linux box running GW6c from just allowing everything in and out? I rely on my router to selectively allow ports in, but im feeling exposed with IPV6 :)

Esentially if I could configure the IPV6 firewall or GW6C to forbid inbound initiated connections for IPV6 would be a good start I think.

I must say this GW6C/radvd is pure goodness. Upon boot up my Vista laptop was issued a IPV6 address via wireless!

Any help apreciated!

edit: The end game being to put my weather station output on the net via Apache. Worlds first ipv6 visible weather stats maybe?

Wasn't 'Node allowed to mirror the gateway installer file?

staogswolbloS writes...

Yeah I just did that but still says my ip is 10.0.0.1.

I meant you can use the ipv6.internode.on.net page to test if you are accessing it via IPv4 or an IPv6 connection (it will tell you whether you are or aren't, and your IPv4 or IPv6 address). Having an interface with an IPv6-enabled interface (with a private IPv4) address is different to actually using an IPv6 tunnel.

I think that ipv6 address it shows in ipconfig is for reference when it is turned on or something.

You'll have an IPv6 interface with an IPv6 address, this will work (and have a working IPv6 address) when you have the tunnel connection set up.

I have trouble accessing IPv6 websites with Firefox (v2.0.0.16 here) but they work fine under IE6 – anyone got any ideas?

Edit: Nevermind, I forgot to close all windows before I tried – once I did that it worked fine (yeah, yeah, I'm stupid :)

Ozzlander writes...

I have trouble accessing IPv6 websites with Firefox (v2.0.0.16 here) but they work fine under IE6 – anyone got any ideas?

Some versions of Firefox require you to enable IPv6 – goto about:config and search for "IPv6" and enable it.

Cool good job internode. Very tempting to transfer now.

Are there future plans to roll out state based tunnels for customers not in Adelaide to connect to?

xghsgd writes...

Are there future plans to roll out state based tunnels for customers not in Adelaide to connect to?

The next step is just to enable it on our LNSes so that your native ADSL service has both IPv4 and IPv6. But that's a big step.

Matthew Moyle-Croft writes...

The next step is just to enable it on our LNSes so that your native ADSL service has both IPv4 and IPv6. But that's a big step.

Awesome. Will look forward to it.

Matthew Moyle-Croft writes...

The next step is just to enable it on our LNSes so that your native ADSL service has both IPv4 and IPv6. But that's a big step.

I'm not overly familiar with PPP. Would this be implemented via 2 addresses on the one PPPoE session, or 2 separate PPPoE sessions?

Works like a charm! Wasn't working at first on XP SP3, but downloaded the BETA version of Gateway6 and it works.

:)

yob writes...

I'm not overly familiar with PPP. Would this be implemented via 2 addresses on the one PPPoE session, or 2 separate PPPoE sessions?

PPP is layer 2. IPv4 and IPv6 are both layer 3, and sit fine on top of a single PPP session.

waagu writes...

Wasn't working at first on XP SP3, but downloaded the BETA version of Gateway6 and it works.

Yep, found the same thing.

Matt writes...

PPP is layer 2. IPv4 and IPv6 are both layer 3, and sit fine on top of a single PPP session.

Oh, of course. I feel a little silly :)

thanks.

Hi,

This is fantastic to see! I've been fiddling with IPv6 for a while – the main attractor to me is that I can eventually do away with NAT. Unfortunately it's been a bit of a chicken and egg situation – the network isn't there because there's no content – and the content isn't there because there's no network to carry it. A bold move by Internode, but I had always suspected that you guys would be the first to offer IPv6 connectivity en masse.

(it's four pages of forum posts and not one traceroute! this is very unwhirlpoolian so here we go for an extra special edition of traceroute6):

traceroute to mirror.ipv6.internode.on.net (2001:44b8:8020:7a80::20) from 2001:388:608c:88b::222, 30 hops max, 16 byte packets

2 clay0-gw-t2-4.net.monash.edu (2001:388:608c:24::1) 0.454 ms 0.245 ms 0.24 ms
3 monash1-gw-v591.net.monash.edu (2001:388:608c:64::1) 0.376 ms 0.274 ms 0.243 ms
4 gigabitethernet1.er1.monash.cpe.aarnet.net.au (2001:388:1:3009::1) 20.163 ms 0.394 ms 0.237 ms
5 ge-1-0-6.bb1.a.mel.aarnet.net.au (2001:388:1:3008:212:1eff:fe56:5a04) 0.762 ms 0.792 ms 0.765 ms
6 so-2-0-0.bb1.a.adl.aarnet.net.au (2001:388:1:6::1) 9.812 ms 9.836 ms 9.802 ms
7 gigabitethernet2.pe1.b.adl.aarnet.net.au (2001:388:1:2004:20f:f7ff:fe2f:e302) 9.808 ms 9.847 ms 9.796 ms
8 gi0-2-35.bdr1.adl6.internode.on.net (2001:44b8:8060:5::4739:1) 10.319 ms 9.98 ms 9.937 ms
9 gig1-17.cor1.adl6.internode.on.net (2001:44b8:8060:3::2) 22.388 ms 22.292 ms 22.649 ms
10 mirror.internode.on.net (2001:44b8:8020:7a80::20) 22.512 ms 22.153 ms 22.131 ms

Traffic seems to take a slightly different path for v6 vs. v4 – is this something we are likely to see in the future with dual-stack network design?

I'm trying to compile gw6c [on Ubuntu] and this is what I'm ending up with.

$ make target=linux all
mkdir -p bin
mkdir -p objs
Building gw6c-config module in ../gw6c-config/
make[1]: Entering directory `/home/david/Desktop/gw6c-5_1-RELEASE-src/gw6c-config'
make[2]: Entering directory `/home/david/Desktop/gw6c-5_1-RELEASE-src/gw6c-config'
mkdir -p objects
g++ -c -I. -O2 -Wall -DNDEBUG -o objects/namevalueparser.o src/namevalueparser.cc
src/namevalueparser.cc: In member function ‘virtual bool gw6cconfig::NameValueParser::ReadConfigurationData(const std::string&)’:
src/namevalueparser.cc:136: error: ‘memset’ was not declared in this scope
src/namevalueparser.cc:177: error: ‘strtok’ was not declared in this scope
src/namevalueparser.cc: In member function ‘virtual bool gw6cconfig::NameValueParser::WriteConfigurationData(const std::string&)’:
src/namevalueparser.cc:259: error: ‘strlen’ was not declared in this scope
src/namevalueparser.cc:259: error: ‘strncmp’ was not declared in this scope
make[2]: *** [objects/namevalueparser.o] Error 1
make[2]: Leaving directory `/home/david/Desktop/gw6c-5_1-RELEASE-src/gw6c-config'
make[1]: *** [export] Error 2
make[1]: Leaving directory `/home/david/Desktop/gw6c-5_1-RELEASE-src/gw6c-config'
make: *** [check-gw6cconfig] Error 2

Can anyone offer up any help/advice on this one?

Tristan Gulyas (Zardoz) writes...

Traffic seems to take a slightly different path for v6 vs. v4 – is this something we are likely to see in the future with dual-stack network design?

No – we're only doing dual stack with AARNET in ADL, BNE, SYD at the moment. No reason other than we haven't quite gotten to it yet. (We turned up ADL and SYD early on when we were turning up IPv6 and just haven't quite gotten to doing the rest – probably now we've released IPv6 connectivity to more people I'll get around to doing it).

Here's mine, using the gateway :)
1 42 ms 42 ms 44 ms 2001:44b8:61::62
2 43 ms 139 ms 44 ms vl67.cor1.adl6.internode.on.net [2001:44b8:
60:8000::1]
3 43 ms 44 ms 44 ms mirror.internode.on.net [2001:44b8:8020:7a8
:20]

Trace complete.

FTTH Test Case writes...

Does anyone know where to look for ideas on how to tighten up a Linux box running GW6c from just allowing everything in and out? I rely on my router to selectively allow ports in, but im feeling exposed with IPV6 :)

You can set up iptables to match and accept/deny IPv6 packets in pretty much the same way that you can accept/deny IPv4 packets.

Google for iptables ipv6 for more info.

– mark

Is the current /60 allocation of address to each routing end point likely to be the norm in the future? Seems a good compromise amount, /48 always struck me as excessive for typical home users. /60 is a nice boundary and leaves one with plenty of options :D

Beldin writes...

Is the current /60 allocation of address to each routing end point likely to be the norm in the future?

Probably not. The freedom to alter policies like that is one of the reasons why I suggested that people not get too emotionally attached to their address allocations upthread :-)

IETF guidelines are for /64's for the kind of people who only ever have one subnet (e.g., the folks we allocate /29's and /30's to with IPv4) and /48 for everyone else. There's a third class of allocation which is not often spoken about because it makes the zealots upset, which is /56, which I suspect will be fairly common (allocated to the kind of people who'd get a /24 in the IPv4 world, or something).

But it's early days yet. None of this is to be construed as "policy."

– mark

I am still getting connecting then disconnecting messages as from log:
22:08:59 Gateway6 Client v6.0-BETA4 build Jun 6 2008-09:15:37 32-bit
22:08:59 Running Windows version 5.1 Service Pack 3.
22:09:08 Your IPv6 address is 2001:44b8:0061:0000:0000:0000:0000:0033.
22:09:08 Using previous keepalive implementation for Windows XP.
22:09:46 General timeout detected.
22:09:46 Disconnected. Retrying.

Any ideas?

Fair Enough, more or less what I thought :)

Agreeing with others, this is a really nice step forward. Looking forward to the native when it arrives.

Trickynik writes...

Any ideas?

Yep, use the beta version 6

Yup that what I am using and I found the problem, it seems my F-Secure internet security is upsetting it. Now the next question, how I allow it..think I have to add the new local area connection to let it though!

Cant wait for all those Google servers sitting in Internode Adelaide to appear with IPV6 addresses :)

ipv6.google.com.au at 40ms latency....mmmm....

FTTH Test Case writes...

Cant wait for all those Google servers sitting in Internode Adelaide to appear with IPV6 addresses :)

ipv6.google.com.au at 40ms latency....mmmm....

.. and it'll look... exactly the same! :)

Nah, the IPv6 version of the google logo animates :-)

( http://ipv6.google.com , for those using the tunnel broker already )

– mark

Seems very snappy, or is it me?

Mark Newton writes...

Nah, the IPv6 version of the google logo animates :-)

Got me on a technicality :)

Mark Newton writes...

Nah, the IPv6 version of the google logo animates :-)

What happens when you have 1 linux computer and 2 windows computers ? Would a possible game of 2 up on Anzac day solve the mess ?

I'd use the Linux computer as an IPv6 gateway. Run the gw6c client and rtadvd under Linux, and put a /64 from your /60 onto your Ethernet, and set all the boxes up to do SLAAP autoconfiguration. You won't even need to assign IPv6 addresses or IPv6 default gateways to the 'doze boxes, it should "just work."

– mark

Thanks Mark.

Mark Newton writes...

You can set up iptables to match and accept/deny IPv6 packets in pretty much the same way that you can accept/deny IPv4 packets.

Google for iptables ipv6 for more info.

And "man ip6tables" is surprisingly useless :-)

It can be a bit hard to get started with ip6tables, but there are sample rulesets out there. An even better start, though, is simply to block everything and start opening stuff up as you find you need it. Or convert an iptables ruleset thatyou know well. You learn a lot that way, plus sample rulesets can be wrong for your particular situation in ways that are very far from useful and – especially when you are just starting out – very difficult to detect or fix.

Tools are generally still catching up with IPv6, so you'll probably have to roll your own filters for a while yet. Another reason to keep it simple.

BTW, it's really important when writing packet filters to remember that as far as your operating system is concerned, IPv6 and IPv4 are totally separate protocols. It's like having TCP/IP and Appletalk on the same interface. Filtering one will have no effect on the other.

Always Young writes...

'm trying to compile gw6c [on Ubuntu] and this is what I'm ending up with.

Jason White posted a patch to gw6c in the ipv6tech list recently. The patch adds a couple of header files to a few of the gw6c sources, and from the look of your error messages, they could be what you are missing.

In src/namevalueparser.cc, use a text editor to add this line below the other header includes:

#include <cstring>

Try compiling again, and if the compilation errors go away for that file, repeat the treatment for the other problematical source files. For source files where some of the errors disappear, but not all of them, try adding this line as well:

#include <stdlib.h>

According to Jason, this issue afflicts people with "a fairly recent version of Linux that uses gcc 4.3 and the corresponding C++ libraries".

Disclaimer: I have not tried this personally, it just looks roughly right.

staogswolbloS writes...

I think that ipv6 address it shows in ipconfig is for reference when it is turned on or something.

Um, this may be way off base, but just in case you didn't know this:

If you enable IPv6 on an interface, that interface will immediately autoconfigure itself with a link-local address – a special prefix plus the interface's MAC address munged in there. With that address, it can then talk to any other IPv6 devices on the same subnet, because they will have done the same thing. Most importantly, it can talk to a router to arrange a proper (routable) address.

For example, one of my interfaces with MAC 00:0E:35:A7:5E:98 gives itself the IPv6 address fe80::20e:35ff:fea7:5e98/64.

If you didn't have IPv6 enabled before, installing the client will have enabled it, hence new addresses will be appearing on your interfaces. However, they don't really have anything to do with the client – the interfaces are doing it for themselves, and those addresses will appear whether a tunnel is up or not.

Looks good, but I dont understand the whole IPv6 fully as yet.
Will this IPv6 make my p0rn download any faster? :P

SnipSnap! writes...

Looks good, but I dont understand the whole IPv6 fully as yet.

Think of it this way. If you didn't have IPv4 addresses on your network (let's say you only ran IPX), you wouldn't be able to access the Internet as we know it unless you went through some protocol translator/proxy.

Over time, if you don't have IPv6 addresses on your network, there will be more and more servers on the Internet that you cannot access.

Noone really knows when this will reach a point where it starts to become a problem, so the general idea is that if you can add IPv6 into your network now, you shouldn't have any issues in the future.

Ideally, you simply won't know or care if the other server you're accessing is running IPv4 or IPv6.

Simon Hackett writes...

Today Internode is making IPv6 services available to its national customer base.

Congratulations on this achievement, and to everyone at Internode who made it possible.

Andrew-ACT writes...

so the general idea is that if you can add IPv6 into your network now, you shouldn't have any issues in the future.

or can wait for a the CPE products to be delivered that do all the work, and take all the pain of this out of it.

White Red Dog writes...

or can wait for a the CPE products to be delivered that do all the work, and take all the pain of this out of it.

Would it not just be an update of the modems firmware?

SnipSnap! writes...

Would it not just be an update of the modems firmware?

you would hope so, but there is no revenue driver to make producers do that, just like drivers for printers & scanners to work in windows vista.

Karl Auer writes...

According to Jason, this issue afflicts people with "a fairly recent version of Linux that uses gcc 4.3 and the corresponding C++ libraries".

Disclaimer: I have not tried this personally, it just looks roughly right.

Thanks. That was on a test box running the latest 8.10 Alpha2, so it would make sense I guess! ;)

I'll give it a go and see if I have any luck!

Thanks for the help ...

Andrew-ACT writes...

Over time, if you don't have IPv6 addresses on your network, there will be more and more servers on the Internet that you cannot access.

But just to clarify for a dummy like me:

I plug my modem into the wall and acc