[arin-ppml] A modest proposal for IPv6 address allocations

[bmanning at vacation.karoshi.com]

 again Michael, please reconsider yoru bad habit of removing attribution.
 do that, and I might reconsider top posting.
 


On Fri, Jun 05, 2009 at 11:19:36AM +0100, michael.dillon at bt.com wrote:
> > 	why do you think that address consumption is related to
> > 	the number of people?  as a referent, check out the 
> > 	"Internet of Things" ((IOT)) stuff.
> 
> The number of end-sites is roughly related to the number
> of people, i.e. with a population of 5 billion, there are
> unlikely to be more than 5 billion endsites. 

	you keep saying that.  on what basis to you make that 
	claim. 


> At any given
> site there could be hundreds of thousands of devices/things
> but that won't consume any more prefixes than a single PC.

	and your reason for this claim is....


> We don't count IPv6 addresses, we count subnets and
> endsite allocations.

	ok... "We" count IPv6 prefixes.  like the prefix
	that is only the top 64bits of the total 128bit
	address.  Or are you coming up with a novel definition
	of "prefix" here

> Anything like a business premises or a cellphone tower
> is a piece of infrastructure that serves a large number
> of people, therefore it is hard to see how the number 
> of these sites could be any significant proportion of 
> the total population. 

	assume.. as a thought experiment, that you have a car.
	the car has the brand - YUGO -...  but I'd be willing to
	wager that the YUGO car company outsources most of the 
	parts production to others.  As an assembled thing, your
	YUGO might get an IPv6 prefix assigned to it so that 
	your car can talk to the road sensors, toll road counters,
	the police vehicles etc.  are you ok with that?

	Now Bill Manning also says (remember to include attribution
	in your replies...) that it is likely that the SLIPSHOD
	brake company has is brake subassemblies in your YUGO.
	It has no desire or requirement to talk to the rest of 
	your car, but it does want to track where all its brake
	subassemblies  are, when the inevitable recall occurs.

	So the SLIPSHOD brake company has given its subassemblies 
	their own IPv6 addresses/prefix to each subassembly.  Hum...
	What just happened?  The car unit and its various components
	have multiple prefixes assigned, perhaps based on manufacturing
	sources.   As you decompose the car into its assemblies and 
	subassemblies, you just might find discontigious and overlaping
	IPv6 subnets.

	Bill Manning sez (remember to include attribution in your
	replies) that this is just one possible example of how the
	IOT is only marginally related/associated with the number
	of humans.

> 
> --Michael Dillon

--bill (who leaves Michaels attributions in place)