IPv6 news

Tue Oct 18 15:42:56 UTC 2005

Michael,

On Oct 17, 2005, at 6:17 AM, Michael.Dillon at btradianz.com wrote:
>> Is VJ compression considered a violation of the "end-to-end"  
>> principle?
> VJ compression happens in the middle of the network, between
> two routers/gateways. End-to-end refers to the hosts, i.e.
> the computers which "host" the end users' applications.

It was a rhetorical question.  My point was that what happens between  
the ends is irrelevant if what gets sent from one end is what is  
received by the other end.  Yes, it is obvious, however I have seen  
people freak out when you suggest touching the address fields,  
regardless of the fact that you say it'll be put back before it hits  
the destination end host.

> Theoretically, in a network, a router/gateway could
> have some intelligence/state so that it does not simply
> forward packets based on destination addresses in the routing
> table. Instead it does some kind of query/lookup to identify
> the real destination location.

Yes.

All of this is simply hackery to get around a fundamental flaw in the  
Internet Protocol (either v4 or v6) - the lack of locator/identifier  
separation.  My concerns with shim6 aren't that the protocol is  
broken, but rather it is so complex that I fear (a) it will take a  
very long time to implement, (b) it will take much, much longer to  
implement correctly, (c) it will never get fully deployed.  Since I  
see multi-homing/renumbering/mobility (all facets of the same thing)  
as the underlying problem with IPv4, I'm hoping that by addressing  
that problem, IPv6 could actually justify its existence in a business  
sense.  Since non-shim6 enabled stacks are already being deployed, I  
suspect an edge box approach will be the most pragmatic way of  
actually getting something people can use.  Unfortunately, delays in  
deploying some sort of multi-homing/renumbering/mobility solution  
will, I suspect, entrench (single sided) NAT even more than it is  
entrenched today, even on IPv6 sites.  So it goes.

Rgds,
-drc