Re: Repository services and budget

[Simon Higgs <simon@higgs.com>]

At 2:00 PM -0500 1/14/97, Perry E. Metzger dug himself in an even deeper hole:

> Vince Wolodkin writes:
> > The numbers below look pretty reasonable.  The only one I would question
> > is the NOC support people.  In my experience running a computer center,
> > if you want three people you either need to hire 5 or contract with a
> > organization who will ensure that you have a qualified person on your
> > site every shift every day.  You will also have to cover 7 days a week
> > so I might double your number.  By my figures 24 x 7 at roughly $25/hour
> > for qualified personnel through a contracting outfit will cost about
> > $240,000, bumping the grand total to just over $1.5 million.
>
> Its far from clear you need that many support people.
>

It is far from clear that you understand what you are talking about.

> The central machines will work fine on their own 99.9% of the
> time. The only time anyone needs to contact the people running those
> machines is if they go down. No one gets access without being part of
> core, you know.
>

And the whole of CORE is down for how long? Hours? That's unnacceptable
(as well as being avoidable and plain dumb). On site response can be in
minutes.

> If you really need 24x7 coverage, an answering service and pager will
> probably do it for you just fine. Monitoring software can take care of
> alerting the operations folks if there is a problem.
>

That solution might work for some of your clients under a tight budget,
but it's *NOT* how mission critical data is handled. Period.

> Furthermore, many organizations contracting to do the work will likely
> have existing 24x7 operations staffs who can deal with the
> problem. Remember that these machines will involve essentially no work
> other than backups on an ordinary basis. You care about the marginal
> work imposed on such a staff -- you don't need the 24 hour staff just
> for this. They'd end up spending all day and all night playing cards
> or watching videos. There would be nothing for them to do.
>

And that's precisely what is required. It's just like the fire
department. If the engineers are watching TV or whatever, then the
system is up and everyone is happy. Only an idiot with an MBA would not
understand the value of this. The royalty revenue lost to CORE during
an outage justifies this kind of redundancy.

> The price numbers I've seen, which keep rocketing up, seem way
> overinflated. This is not a multimillion dollar operation.
>

Then neither should other repositories (like the InterNIC). Uh... the
royalty revenue from registrars to CORE will be millions. Or did you
just forget that small item?

[sigh]

Regards,

Simon

--
If at first you don't succeed, skydiving is not for you.