[25] in SIPB IPv6
Re: IPv6 coordination
daemon@ATHENA.MIT.EDU (Ken Raeburn)
Thu Jul 11 14:55:27 2002
To: Noah Meyerhans <noahm@lcs.MIT.EDU>
Cc: sipbv6@MIT.EDU
From: Ken Raeburn <raeburn@MIT.EDU>
Date: Thu, 11 Jul 2002 14:55:22 -0400
In-Reply-To: <20020711171407.GL9051@locust.lcs.mit.edu> (Noah Meyerhans's
message of "Thu, 11 Jul 2002 13:14:07 -0400")
Noah Meyerhans <noahm@lcs.mit.edu> writes:
> Hello. LCS is currently looking in to IPv6 connectivity, and we'd like
> to coordinate with SIPB to whatever degree makes sense. We've currently
> got a router doing 6-to-4 translation with a machine in... Sweden, IIRC
> on the other end of the tunnel. Better connectivity would certainly be
> nice. Currently at this point I think we'd try getting a tunnel to
> Abeline, and eventually get a pTLA of our own. However, we can try
> something else if we could coordinate something with you. What is the
> state of your IPv6 deployment? What type of coordination between us
> would work for you?
Bill Sommerfeld is the main person running SIPB's IPv6 setup, and the
one who set it up in the first place. He'd be the best to answer
about future plans and any talks that might've been had with the
Network group. I don't know if he's expecting to transfer the
existing setup to them if and when MIT starts supporting IPv6, or if
we'd get a new set of allocations and renumber everything. If the
current setup is at all likely to transition to full campus production
use, I don't think a separate setup for LCS would make much sense
unless you have a separate connection to the net.
But as for our current state....
Our web pages at http://www.mit.edu/sipb/projects/ipv6/ describe our
setup; the status page says:
Our connection is via several IPv6-over-IPv4 tunnels from
limekiller.mit.edu to several different 6bone backbone sites:
* MERIT.
* VIAGENIE.
* VBNS.
We are currently using an address block allocated from MERIT:
3ffe:1ce1::/32. We may renumber at any time; this is the way of
IPv6.
Looking at our router setup, that allocation has been broken down into
a few chunks:
3ffe:1ce1:0000::/48 - main campus; tunnels to people with MIT ties
3ffe:1ce1:0001::/48 - AI
3ffe:1ce1:0002::/48 - LCS
3ffe:1ce1:010x::/48 - tunnels to a couple outside people
While we might want to change that breakdown at some point, I doubt we
need anything bigger than the /32 we have now. Why would you want a
full pTLA?
The 3ffe:1ce1:2::/48 traffic is sent via a tunnel from 18.187.1.231
(limekiller's secondary interface) to 18.26.4.123
(frenulum.lcs.mit.edu), though no contact email address is listed. So
you could just contact whoever's running frenulum and get tunnels set
up from there, or just send outgoing traffic to non-6to4 destinations
through it if you prefer sticking with 6to4.
It appears that we don't have a DNS NS delegation to LCS for that
address block though. If you're actually using that block, and want
to provide PTR records for it, let me know what nameservers should be
listed and I'll set it up.
I noticed a couple things out of date on those web pages. In
particular, the "recent vintage of NetBSD-current" predates the 1.5
release. And the 6to4 gateway I was running has been offline for a
little while. (It was a scratch machine we needed to send to a
conference.) I figured when limekiller gets updated to a newer NetBSD
release we could just do 6to4 on it directly and add the prefixes to
tunnels for people who wanted it. But it also occurred to me that any
changes in the 6to4 setup require changes for anyone we tunnel 6to4
traffic to as a downstream site, so I'm not too anxious to set it up
again except on local subnets, unless or until we have a means to
making such changes happen smoothly, or we can be confident that the
changes won't be needed (e.g., Network group takes over and the "real"
router provides 6to4 service using its fixed address).
Ken
