20000331: Info on hardware for LDM ingestion and decoding (cont.)

>From: "Neil R. Smith" <address@hidden>
>Organization: Dept. Atmospheric Science, TAMU
>Keywords: 200003202042.NAA27616 Unidata platforms 

re: One additional note, DEC's support of Java is not at the forefront.

>What does 'not at the forefront' mean? DEC's web page says they support
>java. ( http://www.digital.com/java/ )

They don't support Java 3D.  We are doing a lot of development work
using VisAD, and that makes good use of Java 3D for volume visualizations.

>What other info is your comment 
>based on?

We try to keep abrest of the kind of support that companies are providing
for the tools we think are going to be very important to our sites in
the future.  DEC (Compaq) is not among the leaders in Java support.

>That would be real sad if the hottest performance/dollar 
>box on the current market that might be needed to do your 3D applet 
>stuff in a metr. synoptic lab (multiple students doing exercises at 
>the same time) wouldn't be purchased because of something like that.

Something like what: not supporting Java 3D?

re: What about things like compiler cost?

>Well, pretty cheap, what with a university site license.

OK, this is good if the university has a site license.  It is not so
good if a department is going after the box by itself.

re: PC ingesting all 4 NOAAPORT, etc.

>And what is that machine/configuration?

The PC we are using is similar to the specs I sent in my previous email,
but it has:

o slower CPUs (450 Mhz instead of 700 Mhz)
o less RAM (500 MB instead of 768 MB)
o less disk (18 GB IBM Ultra SCSIs instead of 36 GB disks)

re: NEXrads

>So, we actually can't currently define a platform configuration that
>anticipates the NEXrad additions because the magnitude of the addition
>hasn't been deciced yet, our local Internet bandwidth may not handle 
>it either, and a queue fix may or may not be available by then anyway.

The queue fix that is in testing here at the UPC will _definitely_ be
available before the NIDS data in NOAAPORT becomes free.

>That's more or less what I infer?  Am I off base?

I am sure that you realize that what looks like a fantastic platform
today, will look like a slowpoke a mere 18 months down the road.
This does not mean that we don't recommend moving up computing wise.
What we do try and say is that it unwise for sites to try and upgrade
all of its computing resources at the same time.  Several sites have
done this, and they are stuck with labs full of machines that are
huffing and puffing along.  It seems more reasonable to upgrade
machines in a more staged manner.  Purchase less machines than you
need next year today, and then purchase newer, faster ones next year.
The problem, of course, is in convincing "management" (deans, etc.)
that this is more cost effective in the long run.

re: Have you checked into higher end PC plaforms?

>We have a preference for brand UNIX boxes and their OS's because of
>our preference for available university site-contract support of our
>mission-critical (teaching/research) activities.  

Fair enough.  That is the other thing that we routinely tell sites.
Their choice of computing platforms should necessarily be swayed by
the discounts and support they can get from vendors that have made
deals with the university.

>Thanks for this responce.  It's really helped. -Neil

I'm glad that it helped some.

>-- 
>Neil R. Smith, Comp. Sys. Mngr.                address@hidden
>Dept. Atmospheric Sci., Texas A&M Univ.        979/845-6272 FAX:979/862-4466

Tom Yoksas