Rebuttal to Mr. Weisberg's insinuations

[Robert T. Nelson]

On Thu, 17 Jul 1997, Peter Veeck wrote:

[snipped beginning of discussion to save space]

> 
> Throughout history there have been two competing beliefs.  Single --
> versus many.  I happen to believe in the philosophy of many.  I don't
> feel that it is good to put all of the eggs in one basket especially, in
> any mission critical situation.  This means that I prefer:
> 	multiple smaller computers over one big computer.
> 	multiple administrators rather than a single administrator.
> 	multiple registries rather than a single registry.
> 	multiple suppliers rather than a single source.

As a general rule I agree with you. Redundancy is important in operational
systems. 

> I feel that there is ample evidence that when you have one organization
> providing all of the information for the root-servers you might reduce
> the number of failures but you will increase the magnitude of the
> failures that occur.  The same is true of root-servers, exchange points,
> backbone providers, and (continuing down the chain) my desk.  
> 
> If there is no way to distribute the functions of a single element or
> organization, then I want checks and balances on that element or
> organization.  If I am dependant upon a single item, I WANT A "SAY" ON
> IT.  In other words, if a single organization is to "control" the
> Internet or any vital part of it, I feel that everybody involved, from
> Grandma at her computer to the highest official at the biggest provider
> and beyond, should have a say in selecting the decision makers for that
> organization.
> 
> I can see no reasonable way to have an Internet wide vote on the
> selection of ARIN's board members who will control IP addresses vital to
> operation.  Therefore one of the solutions that I see is for there to be
> multiple registries.  I see a hand picked registry as only an extension
> of the existing registry.  NAIR is an opportunity for a competing
> registry.
> 

Absolutely agreed. Why don't you become a member of ARIN. I think you'd be
well advised to take a good look at how ARIN can work for you.

I think that the fact of the matter here, is that memory space (IP
addresses) in any programmable system is limited, and at a premium. It
confuses the process if you have more than one entity assigning space in
the Original Place. After that entity assigns space to functions
(downstream registries in charge of how *they* allocate space) they can do
with their space as they please. 

Up till the present day, the Internet has been "under the care of" its
original programmers. In order to have an orderly transition, those
original people have to pass on their functions to a larger, but still
limited group. That group can then, in turn, delegate some of those
functions downstream. ARIN fulfills that function. NAIR could fulfil the
role of downstream registry. 

In the DNS, CORE should fulfil that function, delegating responsibilities
down to individual registries. Just as with IP space, I think that it
would be quite possibly disasterous if we went from 1 provider to a
completely open field operating the root zone. 

In general, people operating networks are not necessarily qualified to do
so, and thus would have problems keeping up with developments - this is
relatively new on the net to have such a proliferation of such networks.

I would prefer that we, as a community, delegate that responsibility to a
group of entities who, collectively, have shown expertise and committment
to development of DNS.  POC and CORE should meet those criteria, just as
ARIN meets that criteria for IP Space allocations. Don't get me wrong. I
can certainly see room for improvement in either process. I don't,
however. think that the proper way to improve the situation is to derail
the process.

If you want a say in IP allocations, join ARIN. Form a registry.
If you want a say in DNS, sign the MoU, and participate. Form a registry.


> They say that a benevolent dictatorship is the most efficient form of
> government.  How do you ensure that the dictator is benevolent?

Jefferson said that you need a revolution, regardless, every 70 years. 
That's 10 years in Internet Time. My bet is that Internet Time will speed
up. We better get planning.

Yours,

Rob Nelson
rnelson at internoc.com