[mesh-dev] Tackling the 10.0.0.0/8 problem

Marc Juul juul at labitat.dk
Sun Feb 8 22:25:04 PST 2015


On Sun, Feb 8, 2015 at 8:50 PM, Mitar <mitar at tnode.com> wrote:

> Hi!
>
> So one thing which is maybe a bit more in the future is that you would
> like to peer with other networks. Often this is done through VPNs:
>
> https://dn42.net/Home
> https://wiki.freifunk.net/IC-VPN
>
> So and the idea is that you can access each others servers and services
> in the network. If you have overlapping IP ranges then that is not
> possible. If you use NAT, it is much harder. How I see it, we are trying
> to build alternative Internet so we should try to achieve end-to-end
> connectivity between our networks. In IPv4 this means that we have to be
> mindful about other subnets.
>
> The whole Guifi.net (27k nodes) is using only:
>
> 10.138.0.0/15; 10.228.0.0/16; 109.69.8.0/21; 5.10.200.0/21
>

We could use a /16 or several /16 networks. If we only allocate a /26 to
each node then we can have ~1024 nodes per /16. This is definitely a way to
go. Since I've never seen a home router use the 172.16 - 172.31 range, we
could use that and get ~15,000 nodes. I think that's a good enough solution
for now.

Alex? Max?


>
> See the imperfect list of what other networks are using here:
>
> http://interop.wlan-si.net/wiki/IPAddressing/List
>

I think using IPv4 for inter-mesh communications is not worth the trouble.
We only need IPv4 to remain backwards-compatible with the legacy internet.
IPv4 should be seen as deprecated for all new internets.

-- 
marc/juul
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://sudoroom.org/lists/private/mesh-dev/attachments/20150208/9d27fea9/attachment.html>


More information about the mesh-dev mailing list