John Hay wrote:
For now what I did in the ptamesh is to use RFC 4193 - Unique Local IPv6 Unicast Addresses. We use that for the mesh itself and also for those that cannot get real IPv6 space. Those that work for Meraka/CSIR get a subnet from the Meraka/CSIR block. Also the inside of our mesh is IPv6 only. If you need IPv4, you need to tunnel it. :-) Part of the reasoning for using the rfc 4193 space was that PI was not available when we started. Also the ptamesh itself does not provide internet connectivity. Just as you need to get IPv4 space from your ISP, you need to get IPv6 space.
We have the same policies of no Internet access within the JAWUG network and we are also currently using RFC 4193 addresses.
However it is my opinion (and this has been voiced by others as well) that RFC 4193 addressing is somewhat pointless. With the relative abundance of v6 addressing resources it seems silly to be numbering your network in such a way that you cannot peer with other networks without the use (potentially) of NAT.
A much better solution would be to obtain global unicast addressing for a network and not announce routes to it. This gives the flexibility to announce the netblock to peers if the need arises without any concerns for addressing conflicts.
Additionally - being restricted to RFC 4193 addressing resources means that that the learning and research that is going on within projects like the Meraka network, the JAWUG network and others is limited to the local network and there is very little scope to be collaborating and sharing with others.
-- Graham Beneke Apolix Internet Services E-Mail/MSN/Jabber: [EMAIL PROTECTED] Skype: grbeneke VoIP: 087-750-5696 Cell: 082-432-1873 http://www.apolix.co.za/ _______________________________________________ rpd mailing list [email protected] https://lists.afrinic.net/mailman/listinfo.cgi/rpd
