This pig is less aerodynamic, and fewer people are pushing. In-addr DNS and whois are simple and well-understood protocols, with many programmer-years of software development behind them.
The problem isn't the marginal cost of a single transaction, that might only be one or two orders of magnitude higher. The problem is the overhead cost of trying to force a poorly-architected system into a semblance of production-quality. If you want something that anyone can _actually rely upon_, that's a precursor to doing the incremental transactions. -Bill > On Dec 4, 2014, at 11:49, "valdis.kletni...@vt.edu" <valdis.kletni...@vt.edu> > wrote: > > On Thu, 04 Dec 2014 11:28:42 -0800, Bill Woodcock said: >>> On Dec 4, 2014, at 11:21 AM, valdis.kletni...@vt.edu wrote: >>> Orders of magnitude? Seriously? I can buy it costs 2x or 3x. >>> But an additional 2 or 3 zeros on the price? > >> Yep, thats why all this is at issue. If it were cheap, and >> worked, like in-addr or whois, there wouldn't be an issue, would >> there? > > So why does an RPKI request cost *500 times* as much as (say) a request > to assign an address block? Why is it *that* much more expensive to handle? >