ARIN is not/is too/is not/is too... blah.

[David R. Conrad]

[note reply-to and cc]

David,

>While everyone is entitled to their opinion, ARIN is no magic
>bullet and is the wrong answer for our industry.  

I don't think anyone is arguing it is a magic bullet.  Whether it is
the right or wrong answer would depend, I guess, on your view of the
role of the US government in an international industry.

>I still assert that there is *nothing* ARIN will give me 
>for my $10,000 per year allocation fee 

Recently, at the APNIC meeting held in Hong Kong, the APNIC membership
voted to modify the APNIC pricing structure, the procedures by which
APNIC allocates the initial block of address space to new ISPs, and
whether or not APNIC should operate a service that could conceivably
compete with services offered by the membership.

How exactly do _you_ influence how InterNIC operates?

>that I don't get right now from 
>the tax dollars I currently pay to support the National Science 
>Foundation.

Your tax dollars are NOT funding address allocation.

> *  It will take money that could have gone to support my network, my
>    employees, and my customers, and instead divert that money to 
>    a yet another bureaucracy.

TANSTAAFL.  Somebody has to pay for registry services.  Right now,
they are being paid for by the domain name charges.  Do you really
want something as critical to your business as address allocations
dependent on NSI given the myriad lawsuits against NSI over domain
issues?

> *  It will increase my costs, which will have to be passed along to
>    my customers, which will effect my business.

Let's look at this a bit (simplifying and not to pick on US.NET, but...):

Size    Fee             Amt of space    Per address per year fee
------------------------------------------------------------------------------
Small   $2500/year      /24 - /19       $9.77 - $0.31
Medium  $5000/year      >/19 - /16      $0.61 - $0.08
Large   $10K/year       >/16 - /14      $0.15 - $0.04
X-Large $20K/year       >/14            $0.08 -> $0.00

So, lets say you have a customer to which you'll be assigning a /22.
Presumably you wouldn't eat the costs if they were a significant
portion of the income you derive from a customer.  Given you indicate
you'd fall in to the "Large" category, this would mean you'll be
passing along a US $3.41 to $12.80 per month cost increase or less
that what you charge for secondarying the customer's domain in the
worst case.  

To be more complete:

                                          % of US.NET's Connect costs*
       monthly "Evil Registry Surcharge"  128K/56K               T1
Prefix     min               max        min     max          min     max
------------------------------------------------------------------------------
/32       $0.0033          $0.0125    0.0011% 0.0042%      0.0003% 0.0013%
/31       $0.0067          $0.0250    0.0023% 0.0085%      0.0007% 0.0025%
/30       $0.01            $0.05      0.0045% 0.0169%      0.0013% 0.0050%
/29       $0.02            $0.10      0.0090% 0.0339%      0.0027% 0.0101%
/28       $0.05            $0.20      0.0181% 0.0678%      0.0054% 0.0201%
/27       $0.10            $0.40      0.0362% 0.1356%      0.0107% 0.0402%
/26       $0.21            $0.80      0.0723% 0.2712%      0.0214% 0.0804%
/25       $0.42            $1.60      0.1446% 0.5424%      0.0429% 0.1608%
/24       $0.85            $3.20      0.2893% 1.0847%      0.0858% 0.3216%
/23       $1.70            $6.40      0.5785% 2.1695%      0.1715% 0.6432%
/22       $3.41           $12.80      1.1571% 4.3390%      0.3430% 1.2864%
/21       $6.82           $25.60      2.3141% 8.6780%      0.6861% 2.5729%
/20      $13.65           $51.20      4.6282% 17.3559%     1.3722% 5.1457%

(*) connect costs taken from http://www.us.net/prices/serv-business.html

Also, while I hesitate to mention it, a possible implication: to avoid
the ERS to those nasty registry people, maybe your customers would
only ask for the amount of address space they NEED?

> *  It will not allow me to increase the size of my current address 
>    allocations any faster than the current InterNIC slow start 
>    policy allows

NOT "InterNIC slow start" -- ALL registries must use slow start and
besides, it was originally implemented at RIPE.

I will note in passing: in the case of APNIC, the membership voted to
change our allocation policy such that it DID directly impact the
amount of address space a new ISP obtains.  Of course, we still have
to abide by the global restrictions of RFC 2050, but the methods by
which the registries follow those restrictions are at the descretion
of the membership.  Presumably this will be the case for ARIN as well.

>This is equivalent to holding a gun to our head and extorting 
>us to pay the $10,000 (or more) annual fee.

Does your electric company extort money from you too?

>the InterNIC
>made 60 million dollars PROFIT last year issuing domain names (while
>funding the assignment of IP address space AT THE SAME TIME).  This
>has to be the biggest money grab in history -- 60 million dollars
>isn't enough for one monopoly to make?  Unbelievable.

You do understand that ARIN is an attempt to make address allocation
independent of NSI and under the discretion of the people who need the
resource ARIN allocates, right?

>The inference here is that by creating a costly new bureaucracy,
>all our problems will go away.  I see absolutely NO evidence of
>any legal or procedural mechanism that will prevent turmoil.  

Please see RIPE-NCC and APNIC.  There doesn't appear to be much
turmoil in either of those organizations.

>There is only one IPv4 address space, so the concept of "alternate
>registries" (aka, like the alternate TLD proposals) has no relevence
>to address space allocation.  Comparing address space to domain
>name allocation is comparing apples to oranges.

No.  The one thing domain name delegation and address allocation have
in common is that they both reply on the Internet community to be
implemented.  If the eDNS crowd ever get a significant following
(e.g., a major service provider takes them seriously), they will be
relevant.  If an AntiNIC becomes established, it would have exactly
the same requirements for relevancy.

>I put an allocation request in last Monday and received my new
>allocation Thursday.  Even if allocation requests could be turned
>around in one-hour, paying an annual $10K fee is not worth it
>to speed the process up three days.  Think about it.

Scenario: NSI loses one of the lawsuits against it.  NSI must pay
damages, etc.  NSI declares bankruptcy.  How long will it take you to
get your IP address space?  Of course, this would never happen.

>There is nothing about ARIN that says we will all be in concensus.
>If anything, there will be tremendous dischord because we will have
>hundreds of ISPs voicing their opinions at the semi-annual ARIN
>meetings.  The current NSF sponsored system does not foster this
>level of turmoil.  If anything, ARIN will turn the currently stable
>IP address policy mechanism into a semi-annual slug fest.

I'm surprised you take such a dim view of democracy and such a
positive view of (arguably enlightened) autocracy.

>There is one -- the same one that has been funded by the NSF since
>the mid 1980's.  Why change something that has worked so well in
>the past?  

The cooperative agreement that created InterNIC (in 1992, not the
mid-80's) expires in '98.  NSF has (in the past) been uninterested in
supporting production services (they are research oriented after all).
As such, it is reasonable to assume they'll not be particularly
interested in continuing in their oversight of the Americas registry
system.

>APNIC and RIPE are not run by governmental entities 

Neither is the registration portion of InterNIC.  Is is operated by a
commercial entity under a cooperative agreement with the NSF.

>They get that address space
>from the current system that is under control of the NSF.  

No.  We get our address space from the IANA, which was funded by DARPA.

>Comparing APNIC and RIPE to the current US model is not fair or
>accurate.

True.  Where APNIC and RIPE members have direct input into how their
registries are operated, American (and South African) ISPs are subject
to the political winds of the US government.  Where APNIC and RIPE
members control how resources are expended, American (and South
African) ISPs must abide by a commercial company's decisions as
(theoretically) moderated by the NSF.  Where APNIC and RIPE members
take responsibility for the administration of the resource on which
they depend, American (and South African) ISPs rely on the US
government to play mommy.

>I believe (as a US citizen) that the Internet is strategic to the
>United States, and control over the address space should remain with
>the US Government.  

And just how does the US government control the address space now?

>Giving control over this strategic asset to a non-profit
>organization that is beholden to nobody is foolishness.

Please see what 501c6 means.  The non-profit organization would be
beholden to the industry it supports.

>Charging for IP addresses will raise the cost of an Internet
>connection.  Raising costs will not improve the health of a growing
>and vibrant industry -- it is anathma to our industry.

I'm surprised you, as a businessman, would have this attitude towards
government intervention in free trade.  Oh, right, as long as it is a
subsidy its OK.

>Lets work together to reduce cost, not increase cost.

And how would you go about doing this, given you have no input as to
how the registry operates?

Regards,
-drc