WAS... Re: pTLA request for RMNET - review closes 23 April 2002

Merlin Merlin" <robert@quantum-radio.net.au
Thu, 11 Apr 2002 17:00:14 +1000


I wonder if I might come in on this conversation for a moment with another perspective.

Regardless of the location of end points, and blocks and bits of blocks it seems to me that the whole idea of moving to
the IPv6 network will die from lack of involvement if it can't become easier to implement. I refer of course to the
actual setting up of the protocols on an actual computer.
While it is of course very necessary to continue working on the outlines - RFCs etc - there needs to be some serious
attempts made to see that valid HOWTOs are produced by those who fully understand the variants.

I take the comment from Pekka Savola in point.
> > > If I was serious about experimenting with IPv6,

Well, there are many people who are serious about experimenting, but the lack of useable information is daunting.
Mailing lists are ok for what they do - but often only confuse the issue. The documents that are available on the
internet now on the subject of V6 are nothing if not conflicting!

The biggest pool of uses or potential users - are of course those already using IPv4. This seems to then be the obvious
starting point to use to build toward eventual take up of full IPv6. That time is of course many many years away. The
investment in training, software, hardware, plant and commerce is so great in the IPv4 area that it will probably never
be fully moved into the IPv6 area in our lifetimes.

As I understand it, 6to4, using the assigned 2002: prefix was designed to enable the use of IPv6 over the existing
infrastructure. An admirable idea, and it appears to work well. However, the depth of documentation on the subject again
is very thin. Enough to get one host or router working if one is lucky, and precious little available to enable a whole
network.
Experimenting? sure. I've been fiddling with it for weeks now on and off. I have one host on my network working as a
host/gateway - finally - I think. and the other host on the network that I set up in the same experimental interest as a
host only is supposed to autoconfigure and connect - well it doesn't. I'm using FreeBSD which seems to be pretty common
throughout the discussions, so it shouldn't be a mystery. But of course it is.

But back to the topic. I've been around the Internet since it was AARNet, so I'm not exactly new to all this. I'm very
sure that if I'm having problems nutting it all out, there is little hope for quite a few others. I know there are
useful things like freenet6 out there, but there again - minimal documentation, and it uses a completly different
prefix, 3fff I think it is from memory. This only serves to further confuse the issue for beginners.

If 6to4 for a number of 'well known platforms' based on the 2002 prefix - designed as I understand it specifically to
use the existing IPv4 networks - could be documented carefully and kept updated it would server to increase interest on
a much wider scale.
I refer to the apparent ease of understanding that numbering system. 2002 is the prefix that tells everyone that it's an
address on an existing IPv4 network and probably is still being used for something useful, like a web server. The next
eight hex-numbers are the IPv4 number translated to hex of the machine that is acting as the IPv6 host/gateway. the
(cb01:6006 in my case) and the ::1 ( I Think) tells that it's the first host on the internal IPv6 network. This is where
it all starts to get grey here. Because the second host - which one would think was numbered ::2 on that network can't
be made to understand that. Any attempt to put that number on any of its interfaces simply confuses it. Interesting
though, both machines can talk to each other via the fe80:<hex-mac-address>:<interface> which of couse is nothing to do
with the 2002 prefix.

Now - I've so far received over a dozen suggestions on how to get the two machines talking to each other correctl, as
well as to the internet, and every one has been different. I have a cardboard carton full of printouts of the same.
Variations of the same theme.

now - I'm not digressing in that discussion above. It's to point out that if it is so hard to set up an IPv6 network
across an existing IPv4 network, using systems supposedly designed to facilitate that, then no one will bother after the
first few frustrating attempts.
If the system isn't loaded too heavily, you should actually be able to connect to http://ruby.chalmers.com.au Apache-2
install page is all, on 2002:cb01:6006::1 Now, I'm not sure if it's actually listening on the v6 port, put a ping6 to
the address should work.
It's the gateway/host/reouter whatever.

s you can see, the origin is the HEX-MAC address of the other host. Which should be 2002:cb01:6006::2 .....OR.... as
someone said, it should be 2002:cb01:6005::1 But of course it would be if it were standalone. But it's supposed to be on
the same network as the 6006 one. You begin to see what I mean.
$ ping6 ruby
PING6(56=40+8+8 bytes) fe80::210:b5ff:fee4:4386%rl0 --> 2002:cb01:6006::1
16 bytes from 2002:cb01:6006::1, icmp_seq=0 hlim=64 time=0.913 ms


So in conclusion - I suspect that very few people actually understand about esoteric details like latency on pure IPv6
machines. But I could point at a user group who I'm sure would love to get their teeth into setting up any number of
hosts, even virtual hosts, behind their one assigned IPv4 address. If someone could come up  with something that was
readable and useable on the subject of setting up 6to4. (and on FreeBSD in my case.) I'm happy to contribute in any way
I can, small as that may appear to the wizards of the pure IPv6 world.

If IPv6 is to be rolled out and not forgotten, people need to be able to implement it on their existing networks.

just my two cents worth,
Robert Chalmers
Quantum Radio





> > > Pekka Savola wrote:
> > > If I was serious about experimenting with IPv6, I sure would not
> > > go overseas to find someone who might be willing to slice off a
> > > part of a block.  However, if I was not serious, I wouldn't care if
> > > all my IPv6 traffic to European countries crossed the
> > Atlantic twice.

> I personally use on day by day basis, IPv6 enabled:
>  - SSH (PuTTY :)
>  - SMTP
>  - Quake 1 + 2*
>  - HTTP
>  - X
>
>

> > guarantee that the 6bone will not be used for production.
> You've got a point there :)
> Though I think most people will profit from good latency.