BBC News - 'Fresh proposals' planned over cyber-monitoring

[Peter Fairbrother]

On 13/05/13 08:46, Ian Batten wrote:
>
> On 12 May 2013, at 15:39, Peter Fairbrother<zenadsl6186 at zen.co.uk>
> wrote:
>
>> On 12/05/13 10:07, Florian Weimer wrote:
>>> * Charles Lindsey:
>>>
>>>> But I thought IPv6 did reserve a batch of numbers that would
>>>> map into the IPv4 space (but not at the "bottom" of the IPv6
>>>> range).
>>>
>>> There are at least three different reserved /96 prefixes for
>>> mapping IPv4 addresses.  Except for the deprecated ::/96 prefix,
>>> these mappings are incompatible with the IPv6 address
>>> architecture and its requirements on the structure of global
>>> unicast addresses, so their use on the IPv6 Internet is not
>>> permitted.
>>
>> I thought the 64:ff9b prefix was routable in IPv6?
>
> Isn't the idea that you use IPv4 embedded in IPv6 until you reach a
> dual-stack machine, and switch out to IPv4 at that point?  There's
> going to have to be some sort of NAT at that stage whatever happens,
> in order for the return packet to get back.

It's not my idea, it's part of IPv6 (which I am complaining about).

>> Mind, I think IPv6 is a horrible kludge with no advantages (apart
>> from the larger address space, which could easily be done with a
>> small extension to IPv4) over IPv4, and should be aborted.
>
> I simply don't understand this argument.  What is IPv6, if not IPv4
> with a small extension for a larger address space (and you say that
> as though it's not terribly important)?

This is of course essential - but is IPv6 the best way to get it? I 
don't think so.

Does IPv6 have any other benefits? I don't think so. There were to be 
some routing benefits, but time has pretty much erased those. Universal 
encryption? Not happening. I can't think of anything else?

> TCP and UDP go over IPv6
> unchanged, for example.  Why would introducing an extension to IPv4
> which would be entirely incompatible (it would require a different
> sized packet header, for example)

IPv4 packet headers are not fixed size - there is a "header length" 
field. It is trivial to add the extra probably 8 or 16 bytes.

> by any easier than introducing
> IPv6?  What's the sticking point in IPv6 which makes it harder?
>>
>> Is the IPv4 240./8 range still reserved? Just start all IPv4e (IPv4
>> extended) addresses with that and make them 10 bytes long, it won't
>> break much.
>
> Aside from every single IPv4 application, router, software stack and
> analyser.  How do you propose, for example, dealing with every piece
> of code that uses sockaddr_in,

Use the 8 normally zero bytes at the end?

Of course, no matter what you do, you will have to make some changes in 
order to get a bigger address space. There is no way around that. It's 
just that with something like IPv4e the total changes are smaller, and 
it's backwards compatible.

> every router that assumes the size of the IP header

I didn't think they were allowed, but I don't know much about backbone 
hardware.

> and every routing table everywhere?

The router looks at the first byte of the "to" address, if it's 240 then 
it uses the new, more logical addition to the table.

(more logical in that eg major hosting companies, ISPs, and other major 
backbone destinations get a single block of IPv4e address beginning with 
240.a.b.c.x.x.x.x.x.x.x.x. The abc bytes are assigned by ICANN or 
someone, the remaining 8 x bytes by the major destination)


I don't claim to know much about internet backbone, or that IPv4e is a 
fully-baked idea - but I don't have any respect for IPv6.

Or HTML5, or RFCs for that matter.

-- Peter Fairbrother


>
> ian
>
>
>