Let us all bend over, apply the Vaseline...

On Wed, 30 Apr 1997, Michael Dillon wrote:

> 
> On Wed, 30 Apr 1997, Bob Atkins wrote:
> 
> > While ARIN will supposedly not be handling IPv6 space I am not
> > convinced that it won't once IPv6 address space comes into use.  With
> > IPv6 ARIN will have an enormous pool of address space. The annual
> > recurring fees generated based on the proposed amounts would be
> > criminal for a non-profit organization.
> 
> The fees charged by ARIN will be under regular review and will be adjusted
> downwards when it becomes clear that they are not needed to run the
> organization.
> 

Again, that is wishfull thinking at best. To be sure the budget
requirements of ARIN will likely grow to absord all available income
and then some. ARIN has absolutely no incentive to lower costs. They
will be a monopoly. Granted one controlled by a

> > Besides my argument is simple - *Who* decided to create ARIN and why
> > isn't such a function put up for bid on a recurring basis?
> 
> Why isn't the Cancer Society or the IEEE put up for bid on a recurring
> basis? Both are non-profit organizations run by their members. Probably
> because the members don't want to do it that way, but if they ever *DID*
> want to do it that way, it could be done. Since ARIN hasn't even started
> operations yet, there is nothing recurring that can be put up for bid.
> 

We don't *have* to rely on the Cancer Society or IEEE. These are
examples of optional organizations. They don't control some crucial
resource. Besides there are many that feel that Cancer Society has
become highly myopic in its course regarding cancer treatment.

> > Consider the way ethernet addresses are doled out. A manufacturer
> > applies for a block - pays a *one time* nominal fee and thats it. 
> 
> The Ethernet MAC address space is considerably larger than the IPv4
> address space and it does not need to interoperate in real time on a
> global basis using routers that are near the limit of their capacity.
> Read up on CIDR and why it is needed to keep the Internet from collapsing.
> 

Yes, I understand. But there is still a need for a management body. ARIN
will not be playing any more significant role than that of handling
Ethernet MAC address space. ARIN will not be functioning as the
routing arbiter. They will only be working to ensure that address
space isn't duplicated.

And yes, of course I understand the issues around CIDR. Are you trying
to tell me that there is some magic voodoo in doling out /19 and
greater (in terms of block size) CIDR blocks? Please don't try and make
this any more complicated - it isn't. The whole thing could be easily
coded an algorithm and implemented in a database frontend.

> > Or
> > consider the way telephone numbers are handled - a one time fee to
> > Bellcore and bang you have 10,000 numbers. 
> 
> ARIN could do this too. The bottom line is that ARIN needs enough income
> in one year to cover its costs. If you think it would be better to jack up
> the fees for address allocations so that the people who get allocations in
> any one year also finance all of ARIN, then you are welcome to make your
> case to the membership. I don't think they would agree that this is a good
> idea, however.
> 

The function of ARIN if handled in a more distributed fashion would be
very minimal. With root servers being provided by large ISPs (based on
their total address assignments), the address assignments handled by
mostly automated systems, the entire manpower for ARIN could be three
fulltime people *at most*. Annual budget would be reduced to well under
$350,000 and the job(s) would be horribly boring. Mainly answering the
occasional phone call since almost all the work would be handled
automatically.

You asked for a sample budget.

We would host a single root server for about $1995/month which would
include the equipment for that price along with a redundant website
service.

In terms of manpower, our hostmaster would probably have the server
admin added to his duties. Server updates would be largely automated so
system monitoring would be the greatest extent of his workload. We
would handle the system with the same high priority that we would
handle any customer's system.

> > The entire concept of leasing
> > IP address space is so open to potential abuse it just isn't funny.
> 
> That's why the practice of leasing or selling address space is frowned
> upon by IANA and by the registries, RIPE, APNIC and ARIN.
> 

Hold on. If $2500/year for a /19 block isn't leasing then I'd like to
know what your definition of leasing is?

> > What is more absurd about all of this is the fact that ARIN would not
> > be in control of the Routing Arbiter's database so in effect they would
> > simply be responsible for in-arpa services and thats all. So how do they
> > intend to enforce their annual fees. Turn off in-arpa services? While
> > that would be inconvenient it certainly wouldn't stop things from
> > working.
> 
> ARIN doesn't intend to enforce anything. It is not a government. It is not
> an army. It is not a police force. ARIN allocates addresses based on a
> policy that the global Internet community has agreed upon. If someone
> doesn't pay their fees that will be public knowledge and it will be up to
> the Internet community to enforce the policy. I do know that the major
> ISPs filter route announcements to make sure that someone is not
> announcing address space that doesn't belong to them.
> 

That is the function of the routing arbiter's database. Entries in that
database are controlled by maintainers. If the maintainer's ID is
valid and properly assiciated with the route objects and those objects
match what is being advertised ia BGP then the routes are accepted.

Unless ARIN is granted some override capabilty they will not have the
ability to influence the routing arbiter's database. I would be
completely opposed to granting anyone overide capability in the routing
arbiter's database, especially an organization like ARIN. All it
would take is one mis-processed invoice and the next thing you find
that you are off the air... And it will take 2 to 3 days to get things
back online. I don't even want to consider the possibilities....

> > outright. As fas as I'm concerned IP space is like radio spectrum. It
> > is a finite resource and it should be managed in such a way as to
> > ensure equitable and fair access to address space to both large and
> > small businesses.  
> 
> That is one of the main goals of ARIN and the other IP registries.
> 
> > I want to clarify that I'm not opposed to paying in some form for
> > universal resources. The problem is that larger more well funded
> > businesses can easily outbid the smaller guys.
> 
> That's why ARIN does not put up address space for auction and does not
> allow ISPs to sell address space to others. A major ISP can get a large
> allocation only because they have agreed to allocate smaller blocks of
> space to the organizations who connect to them.
> 

Again, you are coming from that large ISP point of view. Please read
carefully: Large ISPs are out to dominate the marketplace. They have
no interest in supporting smaller ISPs. So lets not even talk about
that anymore because you obviously do not understand that very
fundamental aspect of this business.

May I ask how long you have been in this business and what your degree
and/or experience is in?

How many businesses have you started? How many people rely on your
business to pay their rent?

My guess is that you have either recently came from a University
environment either as a graduate or as a staff member.

> > However, what
> > I am seeing in both the domain name registery and now the IP regsitery
> > is this concept of centralization instead of diversification. 
> 
> I'm not sure why you see centralization here. There are currently three IP
> registries in the world, RIPE, APNIC and the Internic. After ARIN gets off
> the ground there will still be three registries, RIPE, APNIC and ARIN.
> One of the things ARIN will be doing is helping the South Americans and
> the Africans to spin off their own IP registries to make a total of five
> IP registries, each serving a continent-sized area. So ARIN is actually
> part of a decentralization plan in that respect.
> 

ARIN is the sole management organization for the entire North American
continent. At one time AT&T was the only telephone company we had to
choose from. Yes there was a choice, you could have a phone or not, but
if you did the service came from AT&T.

ARIN is a centralized IP address registery for North America pure and
simple.

> But there's more. Right now the Internic IP registry is basically run by a
> private company (no publicly traded shares) with no possibility for any
> input from ISPs or users of IPv4 address space. By moving it into a
> non-profit organization we will be achieving a decentralization of
> control. Anyone who wishes can become a voting member of this
> organization. Its activities will take place in the public eye and
> mailing lists like this one will be available so that even non-members can
> have some input into ARIN decisions. This is far better than the current
> state of affairs with the Internic.
> 

I'm not saying that the present situation with NSI is any good either
but what I am seeing is just another NSI spinoff. While ARIN would
be an improvement over NSI that doesn't mean that it is that much
better. ARIN is a completely centralized organization. The functions
of IP registery and root server functions need to be separted. What
is left will require far less to support than what is currently
proposed.

If the root servers were put up for bid, several ISPs would be able to
win. Diversification would be assured without the need for additional
staff or to create/purchase that diversified solution independantly.
Overall costs would be lowered and ultimately that is what we need.  My
preferred method would be to require the largest IP block holders to
host a root server at no charge. All ISPs would pay basic annual fee
for address management regardless of how much address space that have
assigned.

Assuming that there are around 600 ISPs that would be doing business
directly with ARIN, an annual fee of $600 each would produce $360,000
in annual revenue and that would be more than enough to cover 3 staff,
facilities, services and equipment. Add a nominal charge of $0.10 per
host for address space and you have a healthy additional income that
could easily bring the annual income to over $500,000. Each paying ISP
would automotaically be a member and so would be able to directly
contribute to ARIN in other ways.

My basic gripe is this: $2500/year for a /19 block is highway robbery.
The costs estimates being made for ARIN clearly are not reasonable
and indicate a fairly bloated budget. I truely believe that it
is possible to provide the *same* level of service for a lot less.

---
===========================================================================
Bob Atkins, President			| bob at digilink.net
Digilink Network Services		| http://www.DigiLink.Net/
Switched ISDN Internet Access		| mailto:info at DigiLink.Net
310-577-9450		"Our business is your network"
===========================================================================
The man who follows the crowd will usually get no further than the
crowd.  The man who walks alone is likely to find himself in places no
one has ever been.
                -- Alan Ashley-Pitt