LET'S JUST GO AROUND

[Jeff Binkley]

HC>At 8:55 AM -0500 2/3/97, Jeff Binkley wrote:

HC>>
HC>>PF>>It's kinda fun listening to the ISP voices saying "we wanna be
HC>>PF>>independent, we wanna be the final authority, we wanna make our
HC>>PF>>own choices without regard for anyone else."

HC>>
HC>>PF>ISP's can already do this; if they can justify PI address space,
HC>>PF>they can decide to go to the InterNIC to obtain it. Once again,
HC>>PF>the fact that it may or may not be routable is an orthogonal
HC>>PF>issue. Under the ARIN proposal, the only functional difference is
HC>>PF>now they will pay for the services rendered by ARIN in obtaining
HC>>PF>address space directly from ARIN.
HC>>
HC>>
HC>>Which brings us back to the whole purpose/benefit of this proposal.
HC>Why >should they be forced to pay for something they don't have to
HC>pay for >today, only to have no/limited perceived benefit ?

HC>But they don't pay for it today because it's been funded by NSF.
HC>They do get benefits of controlled unique addressing.  Addressing of
HC>this sort is a prerequisite for global routability, but they are not
HC>the same thing.

HC>>This whole thing reminds me of the government trying to levy taxes.

HC>Take a different government example.  When I was a kid, and my mother
HC>took me to visit a national park, there were no admission fees or
HC>very small fees.  Jeff, you don't suggest here that there are no
HC>costs to running a Yellowstone or Yosemite, do you?  The point I'm
HC>making is that the park operations were subsidized by general tax
HC>funding.

HC>As federal budgets become tighter, there's been more emphasis on user
HC>fees for services, privatizing services, etc.  Same thing, in my
HC>mind, whether it is the Park Service or NSF.

I don't disagree but general funding means everyone pays unless they 
become tax exempt.  This means folks who use the park or not.  It's like 
where I live we subsidize the schools via property taxes.  My wife and I 
have no kids and my neighbor has 5 yet we pay the same taxes.  That's 
quite a big different than what is being proposed.

HC>>I've watched much of
HC>>the discussion going on here and many of the supporters tend not to
HC>be >ISPs or folks who would be directly finacnially impacted by this
HC>>proposal.

HC>I really think you need to distinguish between the smaller ISPs that
HC>emphasize customer connectivity, and the larger network service
HC>providers (NSP).  Think of the latter as those firms with national or
HC>large regional backbones, whose primary business is moving large
HC>numbers of bytes rather than providing web services, dialup access,
HC>etc.  The latter don't seem to be complaining here, and I think that
HC>is because they take registry services as a cost of doing business.
HC>NSPs have been quite active in the IETF, NANOG, etc., so these
HC>proposals are not a surprise.  NSPs, I suspect, feel they don't need
HC>to complain because they have already gone through the discussions in
HC>the RFC2050 effort, etc.

The later don't seem to be participating here.  That's a big difference.


HC>Don't assume 2050 and related documents were things where it was easy
HC>to reach consensus.  Sprint had pushed for the minimum allocation
HC>being /18, and a lot of effort was to reach compromise on a /19.

HC>For whatever reasons, the smaller ISPs are just starting to get
HC>exposed to some fairly well-developed issues.

HC>>From my unofficial counting the supporters tend to be:  NSI,
HC>>hardware vendors, academic affiliated individuals and a few other
HC>>interested parties.  The opposition/concered parties tend mostly to
HC>be >the ISPs and network providers.

HC>Other than arguments over the details of the proposal -- parts of
HC>which I don't like either -- could you point out a NSP that has major
HC>technical problems with the proposal?  Smaller ISPs, certainly.  Some
HC>of these smaller ISPs also are demanding address portability and
HC>global routability that no one knows how to do in a reliable and
HC>scalable way, regardless of how many people scream "there _ought_ to
HC>be a way to do this."  In fact, address portability is, IMHO, an
HC>obsolescent issue.  Current good practice is to design systems such
HC>that they are easily renumbered, or to translate addresses on a
HC>firewall or gateway that might be installed in any case.

HC>Don't misinterpret what I am saying to mean I am opposed to smaller
HC>ISPs in the market. I think they have a critical role in supporting
HC>end user services, whether dialup access, web hosting, etc.  I'm
HC>writing this from a personal account, and it is no accident that it
HC>is with a local ISP rather than AT&T, MCI, etc.  But the smaller ISPs
HC>simply have not grown up aware of the operational scalability issues
HC>necessary, at least in the short term, to let the Internet grow and
HC>prosper.

I would guess they are painfully aware of some of these issues just 
perhaps not on the same scale.  

HC>>This is akin to the "not in my
HC>>backyard" syndrome of where to build prisons and the like.  We all
HC>agree >they are needed but don't build them next to where I live.
HC>With ARIN is >seems we agree there needs to be some control over
HC>address space (albeit >we would probably disagree on how much control
HC>and what the real purpose >of the control was for) but the supports
HC>are saying make the ISPs pay >for it, while the ISPs are saying wait
HC>a minute.  They weren't even the >ones asking for it from what I can
HC>see.  Paul's point is there will even >be limietd benefit for them,
HC>even if they go along with it.  So why >should they start coughing up
HC>money for something which has this little >potential for them ?

HC>It comes down to a simple question...you agree there is a need for
HC>control "some over address space."  You agree, I belive, that
HC>control providing such involves expenditure of funds.

HC>Then my question to you:  who pays?  I don't really care who pays as
HC>long as the function is funded, and I believe that costs will
HC>eventually be reflected in pricing no matter who is charged.  I also
HC>believe that the costs will be relatively small in relation to the
HC>profits to be made and the benefits perceived by customers.

I won't disagree that the address space needs to be managed.  I believe 
this is first an engineering question and not an organizational/economic 
one.  I think once the engineering pieces are figured out then we can 
move onto the organizational and economic pieces.  I think it would be a 
mistake to use economics to force compliance to engineering rules 
because if so done, technology will soon replace the economic pressures.  
history has proven this.


Jeff Binkley
ASA Network Computing

CMPQwk 1.42 9999