URLs, IPs and interception

[Roland Perry]

In article <47CAD568.1080706@defoam.net>, Adrian Midgley 
<amidgley2@defoam.net> writes

>>>using IP  numbers...
>>
>> <cough> they are *numbers* which represent *IP addresses*
>which are 32 bit/digit binary numbers, no?

Yes, but it's important to get the lingo right. Imagine how asking 
someone for their "telephone address" might grate, well that's what it 
feels like to people in the Internet industry when folks mention "IP 
numbers".

>> But you can't always access a website purely by IP address.
>But with IPv6 there would be a more limited need to share IP numeric
>addresses, and in the subset we are possibly discussing there might be a
>greater desire amongst host operators (perhaps assymetrically at the two
>ends of a connection) to have one machine on one address.

ipv6 is simply "more numbers, therefore more addresses". It may still be 
a useful thing to have several websites served at one address, rather 
than try to have one server box (which would be sorely under-utilised 
serving just one website) having numerous addresses. I'm not sure this 
kind of thing is discussed very much.

>I'm not convinced that the problem as originally presented - misuses by
>ISPs of DNS requests -  actually gives a problem with multiply hosted
>FQDnames on singe IP numeric addresses though.

Sorry but you've lost me there.

>If one has a DNS in one's own network then there is no reason it should
>not know all that, and as many reasons that it should as there are
>anywhere else.
>
>So the DNS lookup within my firewall would return the correct csingle IP
>address, and the packet containing the one of many FQDNs for a
>virtual/server on that address would then be sent in its direction, and
>interpreted as usual, but without any need for a DNS query to escape
>purdah here.
>
>It implies an undernet synchronising the DNS, which is presumably
>absolutely stuffed with obCryptos.

And there :(
-- 
Roland Perry