tmb at maddog.u-net.com
Sun Jun 3 22:48:46 BST 2001
It's increasingly popular for ISPs to use some kind of
RFC1918-addressed IP network internally, even when they assign public
IP addresses to customers. I note that both tele2 and BTinternet do
this at least. This lets them play bizarre routing games or get a
whole lot of addresses all to themselves, however they should be able
to do most things (except service millions of customers concurrently)
without this fairly ugly technique
The key question is whether or not you have an RFC1918 IP address
yourself. If you do, then it makes sense for them to use an RFC1918
address system and NAT your packets for you. However, if you get a
public IP address and they still transit you over an RFC1918 network
then I'd love to understand why myself.
I'd also be interested in knowing if they expose these RFC1918
addresses via ICMP (i.e. traceroute) to other external networks.
The whole point of that RFC is to *hide* private network addresses,
but a traceroute will usually bring it out.
However, I don't do network design and I might be missing something.
> Can anyone direct me to an explanation why one's first hop should
> be to a RFC1918 IP; and am I right to deduce that a packet loss of
> 30 to 40% to this IP and no packet loss to the modem means that all
> is well with the FreeBSD server at my end and the fault is upstream?
> How can I test and log the link performance without contributing to
> its failure?
> It's not working! Slower than 'cuts' sometimes!
> ------ FreeBSD UK Users' Group - Mailing List ------
More information about the Ukfreebsd