[Info-vax] implementing IPv6 on the internet

[Dirk Munk]

Jan-Erik Soderholm wrote:
> Den 2016-09-23 kl. 23:22, skrev Dirk Munk:
>> David Froble wrote:
>>> Dirk Munk wrote:
>>>> Richard Levitte wrote:
>>>>> Den fredag 23 september 2016 kl. 20:59:55 UTC+2 skrev Dirk Munk:
>>>>>> With "keep on dreaming" I was referring to your translation on the CE
>>>>>> router idea, IPv6 over the Internet, IPv4 at your home LAN.
>>>>>
>>>>> Oh, I'm sure someone will love NAT enough to do just that.
>>>>>
>>>>> Cheers,
>>>>> Richard ( #ishouldshutupnow )
>>>>>
>>>>
>>>> It's not that easy. It's not only the addresses, in many cases the way
>>>> the packets are set up has changed as well. Such a translation would
>>>> be quite a difficult undertaking, the CPU power of a simple CE router
>>>> isn't enough.
>>>>
>>>>
>>>
>>> Look, I'm not an expert at this stuff, but I have to ask, why is it so
>>> hard?
>>>
>>> Right now, NAT somehow figures which internal address to send a response
>>> coming back from the internet.  I think I read once that it puts the
>>> internal IP address in the packet.  Don't know much about that.  But if
>>> so, then at least returning packets via IPv6 could have inside the
>>> packet the IPv4 address for the internal system.
>>>
>>> As for connection attempts from outside, the sender would have to have
>>> knowledge of the internal address, and perhaps places that in the
>>> packet.
>>>
>>> Not real clean, but, wouldn't it work?
>>
>>
>> I'll try to explain.
>> suppose my pc wants to make a IPv4 connection with  microsoft.com, what
>> happens?
>> 1. It will do a nslookup for microsoft.com
>> 2. it will get the IPv4 address X4 (and the IPv6 address X6) of
>> microsoft.com
>> 3. it will construct a packet with destination address X4, and sender
>> address LAN4
>> 4. It will send that packet to the router.
>> 5. The router will exchange the LAN4 sending address by its own WAN4
>> address.
>> 6. The packet will be send to microsoft.
>
>> 7. return packets the other way around.
>
> Point 7 is very oversimplificated... :-)

Sure, I know. But the essence is the simple exchange of addresses (and 
port numbers)

>
> The return package arrives at the WAN address (there is only one,
> don't really know what you mean with "WAN4").

The IPv4 address of the WAN port

> The whole "thing" with
> NAT is to decide where to sent the return package. The NAT tables
> holds the originating internal IP address (and port no, since that
> is also dynamicaly created by the web browser for each call).
>
> So in your step 7 is where most of the "NAT magic" happens... :-)
> It is not just "the other way around".

You're right, but that is how it's done, not what is done if you 
understand what I mean. I wanted to bring the whole thing back to the 
bare minimum.

>
>> 5. The router will exchange the LAN4 sending address by its own WAN
> address and also storing the LAN4 address and LAN4 port number in the
> NAT tables and saves this as a WAN->LAN port number cross ref.
>
>> 6. The packet will be send to microsoft.
>
>> 7. Return packets arives to the WAN address of the router at the port
> number the router specifed in step 5 above. The router then knows that
> any traffic comming into that specific port should be routed to the LAN4
> IP addres and LAN4 port number stored in the NAT tables. That line in the
> NAT tables is then cleared out.
>
>>
>> Let's see what happens with your idea.
>> 1. It will do a nslookup for microsoft.com
>> 2. it will get the IPv4 address X4 (and the IPv6 address X6) of
>> microsoft.com
>> 3. it will construct a packet with destination address X4, and sender
>> address LAN4
>> 4. It will send that packet to the router.
>>
>> And now we get a problem. How is the router suppose to know it has to
>> build
>> a new IPv6 packet with the same payload, but destination address X6, and
>> sending address WAN6?
>