Re: Repository services and budget

[Dave Crocker <dcrocker@imc.org>]

Rick,

	What follows, of course, is a vigorous disagreement with your
assessment, but just to make clea,r ahead of entering that fray, that you
are taking a clear and concrete position and I, for one, appreciate that it
permits legitimate and (one hopes) productive exploration of specifics.

At 7:58 PM -0800 1/13/97, Rick H. Wesson wrote:
>Your estimate was for a cripled set of linux boxen, and provides little

	I don't recall specifying linux, though I know others did.  I
allocated some money for o/s platform and some more for database software.
It happens that I suspect a PC-based Unix system (bsdi, linux, solaris) is
just fine, but that's not a position I'm taking religiously and believe
that the budget would be fine for a workstation platform.  The debate, of
course, has to do with the level of demand to be placed on the platform.
On that, it would seem, there is a basic disagreement.

>well work 24 hours a day and 7 daze a week!

	We agree on this.  The assumption that my previous efforts excluded
this requirement are not correct, though of course I do understand that one
might continue to disagree that it is satisfied.

>Ongoing
>===============================================
>3 software developers 	80K each/yr	240,000

	This assumes on-going software development and I don't believe that
that is necessary.  The repository has some very, very basic functionality
to provide; it is not intended as an evolving customer support/service
engine.  THAT is the job of each registar's customer support folks to
provide.

>3 T-1 Connections	12K each/yr	 36,000

	This underscores a very basic point which adds considerably and, I
believe, unnecessarily, to the expenses.  It is based on the "in-house"
model for connectivity, rather than using co-location services.  In my
opinion, the primary access to the repository should be with machines that
are co-located at well-connected points.  The cost of such access is
dramatically lower than your model and the service will tend to be better.

	(By the way, I believe you underestimated the costs of this
connectivity model, since you need to provide both for the telecom wire to
the provider AND for the provider charge.)

>3 Rotateing 24/7
>    NOC support ppl	35K each/yr	105,000

	If on-site staff is required for 24/7, the actual staffing is
typically 4 or 4.5 people.

	But it isn't necessary. Basic operations require no
moment-to-moment intervention, except for periodic system administration.
Problems require intervention by experts.  The operators that one hires for
24x7 coverage do not typically have the skillset for such interventions.
Hence, what IS required is one operator and a damn good monitoring and
alert service to page an expert.  How much time, per year, is required for
such an expert?  I'd guess one FTE, spread across 2-3 senior people.

	On the matter of reliability it's important to remember that this
is a replicated system, so that outage of one of the machines affects
capacity but does not necessarily require instantaneous repair.

>Travel					 80,000
>Telephone				 35,000
>Legal Team				125,000

	While we all are aware of the many threatended lawsuits that have
been cited on this list, I'm not at all clear why the REPOSITORY needs such
a large legal budget.  Please explain.

>
>Startup
>==============================================
>Routers					 25,000

	Not if co-location is used.

>COMPETITIVE services against NSI. Why should it not be done as robustly
>as possable? Afterall these businesses are betting their livelyhood on the

	We don't disagree about the need for robustness.  This is a
difference about means, not ends.

>It does puzzle me as to why the IAHC does not want to discuss the
>functionality
>of the CORE repository, as that functionality defines what and how services

	No idea what 'refusal' you are talking about.

>I would like to add that if the CORE repository is run at a fixed cost,

	No one said it would run at a fixed cost.  That was a conclusion
some folk jumped to because I did a one-off budget, to try to get an
estimate for its initial requirement.

>If all this is to be determined by the CORE I could only say that for the
>first
>time "on the wire" protocols will be defined by a small closed group and not
>by a Working Group process. Is this wize?

	You are assuming that protocols will be invented by CORE.  I don't
know why you assume this, but the assumption differs from my own.

d/


----------------------------
Dave Crocker, Director                                       +1 408 246 8253
Internet Mail Consortium                                 (f) +1 408 249 6205
127 Segré Place                                             dcrocker@imc.org
Santa Cruz, CA  95060 USA                                 http://www.imc.org

[IAHC member, expressing strictly (or loosely) personal opinions]