Re: [Idr] WGLC on draft-ietf-idr-as-private-reservation-00

[Shane Amante <[e-mail]>]

On Nov 30, 2012, at 12:22 PM, John G. Scudder <[e-mail]> wrote:
> On Nov 30, 2012, at 12:34 PM, Chandra Appanna <[e-mail]> wrote:
> 
>> I figure the chairs might want to hear from some of us who are reading but silent so far..
> 
> Indeed. Thanks for speaking up and I encourage others to do the same if you have an opinion on the subject.

FWIW, I support advancing/publishing this document.

While I personally find private ASN's distasteful and I try to avoid them whenever possible (especially/even in private network contexts), the facts of the matter are:
1) The genie is already out of the bottle with the existing range of private ASN's (64512 - 65534).
2) The, <ahem>, "hacks" to deal with them are already in code and would, hopefully, just need a trivial modification to recognize the new range.
3) We operators already have methods of filtering private ASN's and those filters should also require trivial _one-time_ modifications, shortly after this is published.

And, at the end of the day, I would much rather see that the IETF adopt a standard range of "Extended, Private ASN's" than forcing operators who need a large set of Private ASN's to squat on legitimate space.  That would that create a mess.

-shane

P.S. -- Frankly, for folks who don't support this proposal I'd say this is really about economics.  If RIR's supported a model of leasing ASN's in bulk, e.g.: N x 10,000 4-Byte ASN's for $X/year, (where $X was an extremely small fraction of: N ASN's x $500/ASN) for the whole block.  Obviously such bulk allocations would require needs based justification, as defined by RIR policies.  If such a proposal were to be adopted, it would /potentially/ *also* make it more palatable to anycast operators on the Internet to use a globally unique ASN per anycast instance/node, for reasons outlined in this draft:
http://tools.ietf.org/html/draft-mcpherson-unique-origin-as-00
(Not sure if this became an RFC).  IMO, this course of action would be a much more palatable to me, because it would address the needs of both anycast operators and DataCenters/Enterprises.  
   All this really begs the question: when will we recognize that numbers are not a precious commodity, especially ASN's that only appear in the RIB, not the FIB?