Re: [MBONED] addrarch: IANA allocations and assignments

On Wed, Dec 20, 2006 at 01:10:01PM -0800, Toerless Eckert wrote:
> On Wed, Dec 20, 2006 at 07:33:23AM -0800, David Meyer wrote:
> > >     - no approach seems to be needed for this (given SSM and v6 methods),
> > 
> > 	That would, however, seem to be wrong, given the real
> > 	world. SSM deployment seems to be stalled (and "my DSLAM
> > 	doesn't do IGMPv3" [or whatever]), and IPv6 multicast,
> > 	well, nuff said. 
> 
> Which is why if i would have any vote in this, there should be no relaxing
> of static allocations of IPv4 multicast addresses. All cases i have
> seen (and am still presented with in customers) could always use SSM, but
> year after year customers delay it still referring to not being able to use
> SSM because they need to support things like old host OS - instead of
> pushing their vendors anywhere to support IETF standards SSM RFCs (especially
> IGMPv3 of course).

	That is true.

> As soon as there will be easier ways to hand out global IPv4 addresses,
> the already pretty lame checking that the intended purpose is NOT an
> SSM usage will get even more lame and we will get even more SSM applications
> relying on interdomain PIM-SM.

	Well, I'll be more explict. Many of the addresses that
	are assigned to folks *are* SSM applications.

> Either the IETF has an opinion on interdomain SSM vs. interdomain ASM,
> or we don't.

	Who is the IETF, exactly? And what authority do they (if
	you can locate "them") have over what is run on the
	Internet? Real questions that bear on your point.

> It would be fine though to hand out global scope IPv4 addresses to
> applications that should be SSM if:
>   a) There's charging for it (as Marshal explained)
>      (which you have always refused as impractical though, arguing
>       registries are too lame trying to charge money for multicast addresses)

	The IETF, as I understand it, isn't really in the
	business of endorsing business models, so this one is not
	in the IETF domain (again, AFAICT).

>   b) If it is justified by the usage of certain non-SSM compliant
>      softare (eg: PC_OS x.y).
>   c) If it has to be renewed every year.

	I understand your concern and reasoning, but you also
	realize that you just made a bunch of stuff up. That's
	fine, but other people make up other stuff, which is not
	necessarily consisitent with a)-c)

> b) and c) together would ensure deployments using SSM application
> will not abuse global scope ASM forever.
  
	Not sure about that, and BTW, "forever" is a long time. 
  
> It would also actually encourage moving to SSM if the address allocation
> process causes the evaluation this is SSM, and they can have addresses
> as long as they actually have a plan to move towards SSM.

	I got the second part of the sentence, but didn't
	understand the first part.

	--dmm