BBC does IPv6 ;) (Was: large multi-site enterprises and PI

Mon Nov 29 22:49:26 UTC 2004

On 27-nov-04, at 22:45, bmanning at vacation.karoshi.com wrote:

>>> the short version of my rebuttal is: "those are not your bits to
>>> waste."

>> They are if my ISP assigns them to me.  :-)

> 	er... not really.  they are the ISPs.

Well, the ISP doesn't "own" them either. But they're assigned to me, 
which gives me the right to waste them as I see fit within the limits 
of the address assignment policy. (Which allows considerable leeway 
towards bit wasting in IPv6.)

>>> second, let me add, "and it's not your routing table, either."

>> I have no idea what this means.

> 	if you have no idea aobut the impact of address
> 	assignment on routing tables,

I think anyone who has been present during the address policy sessions 
in the last few RIPE meetings can testify to the fact that I certainly 
have ideas about this.

What I mean is that the remark that something is not my routing table 
makes no sense to me. Nobody owns the abstract global routing table. On 
the other hand, obviously I own the memory in my private box that 
happens to have a particular instance of the global routing table in 
it.

>   then you really should
> 	spend some time implementing routing policies -before-
> 	you burn cycles telling others about how they should
> 	run their networks.  no one is stoping you from implementing
> 	whatever prefix acceptance/forwarding policy you may
> 	chose to implemenet for -YOUR- customers. it is a -local- effect.
> 	just stop trying to tell others how to manage their
> 	routign tables.

Unless I'm experiencing blackouts, I haven't been telling people how to 
manage their routing tables. The trouble with the routing table is that 
it's not really manageable: in theory, you can filter out the stuff 
that you don't like, but in practice this can't be done without 
breaking reachability, so we're all forced to live with the sum of all 
crap that anyone feels fit to inject in BGP on some corner of the 
planet. That has been my point all along: we should empower operators 
to make reasonable tradeoffs between optimum path selection and routing 
resource consumption.