<div dir="ltr"><div dir="ltr"><div class="gmail_quote"><div>Hello. I found this project through the TUHS mailing list. I'm interested in old protocols like X.25 and ATM, though I don't actually have any software nor hardware to use them with.</div><div dir="ltr"><br></div><div dir="ltr">On Sun, Sep 2, 2018 at 8:53 PM Michael Kjörling <<a href="mailto:michael@kjorling.se">michael@kjorling.se</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">>> Note: I'm allocating a /48 to each member so that they can divide it<br>
>> into many /64s.<br>
> <br>
> A /48 would be the smallest block I'd want to dish out as well.<br>
<br>
/48 gives plenty of room for subnetting, but is there an appropriate<br>
IPv6 block to carve them out of? It doesn't _really_ matter as long as<br>
nobody cares about global routing on the same host, but could<br>
potentially complicate routing for multihomed hosts unnecessarily. Or<br>
are we willing to rule out use on multihomed systems?<br>
<br>
There's the old, deprecated site-local space, which would provide<br></blockquote><div><br></div><div>Hopefully there isn't some mismatch between old systems actually treating them as site-local unicast with the special rules, and modern systems treating them as global unicast.<br></div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
sufficient room. Properly done RFC 4193 space doesn't; that gives only<br>
a single /64.<br></blockquote><div><br></div><div>RFC 4193 says that "locally assigned" prefixes (i.e. in fd00::/8) have a 40-byte unique ID and give a /48 prefix. So you can assign a /48 per user, there is no problem.</div><div><br></div><div>But RFC 4193 doesn't define any specific method for assigning prefixes in the "centrally assigned" fc00::/8 range. If cjdns can get away with littering the entire /8 with pubkey-based addresses, you could also get away with assigning fc00:<ipv4>:<ipv4>::/48 or fc00:<idx>:<idx>::/48 to each user.</div><br>I would also consider 3ffe::/16 (old 6bone) – unlikely to be reused anytime soon, and with nostalgia already built in.<div><br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
Or ffff:<ipv4>:<ipv4>::/48? But I think that's special-cased by a lot<br>
of software...<br></blockquote><div><br></div><div>All of ff00::/8 is multicast addresses; you really don't want to use that for unicast. (Though it sounds like you're mixing it up with 0:0:0:0:0:ffff::/96, which is another special case.)</div></div><div class="gmail_quote"><br></div>-- <br><div dir="ltr" class="gmail_signature"><div dir="ltr">Mantas Mikulėnas<br></div><br class="gmail-Apple-interchange-newline">(Apparently my presence in this list is a blatant violation of <a href="http://chivanet.org">chivanet.org</a> terms of service. I had hoped that something like RetroNet would have included the old idea of a global internet without country or continent borders, but oh well.)</div></div></div>