ipv6 exploratory virtual appliance

[amcintosh]

I'm putting together an Oracle VM VirtualBox appliance running Lucid Lynx 10.04 (LTS) to distribute to the local high school robotics club on a USB stick in order to explore ipv6.  This is just a cleaning and repackaging of my own learning environment.  I may also pass it out to selected college students in my instrumental chemistry course.  "Pass it out" does not mean "personally absorb all the direct costs."  It will almost certainly be a "Kickstarter" project that covers costs.

I'm thinking...
1.  having a .ssh directory with a technical support key in /etc/skel so all new accounts can get support.
2.  have Apache running with a generic web page discussing ipv6 resources.
3.  have each USB serialized, so something intelligent can be done with the MAC addresses / v6 network prefixes.
4.  have a streamlined process so they can sign into Hurricane Electric, and then some PHP script or other can do their configuration for them.
5.  have the appliance facilitate letting them connect and interact with each other's systems.
6.  have apt-cacher running.
7.  does anyone know a "must upgrade" feature in Precise Pangolin (12.04 LTS)
8.  many of them don't have routable ipv4 addresses.  Provide some sort of peer to peer tunnel in the appliance.

I'm pretty much on schedule for the transition from "a little longer than I thought" and "why did I ever start this."  The web page is occasionally up at http://lifepod6.vima.austin.tx.us/LifePod (though I am now in Morris, MN)

At this time I have something solid enough to tell what it really does, and listen intelligently to the suggestions for things I left out.

So ... speak up and be sure I have a reasonably attainable wish list.

I do have an HE /48 assigned.  The local internet provider is strictly IPV4, although it is a close knit college town, and their technician may be amenable to drinking beer and joining the effort.

Does it make sense to try to make all these separate pods be on /64s cut out from my /48 to facilitate the "group" identity, troubleshooting, or routing experiments?

What tunneling technology should I use for the peer to peer in a mixed routed / non-routed environment?

I assume that having apt-cacher running, and all of them knowing about each other, can pretty much keep the Ubuntu update bandwidth mostly local.

Is there a way, and a reason, for them to have their routing "fail over" to each other?  I think most of us are on the same telephone carrier, and we all have fiber to our house.

Should I turn them all loose toward HE with no coordination?