From mazz@ipac.caltech.eduMon Feb  3 16:15:52 1997
Date: Tue, 3 Dec 1996 16:11:05 -0800 (PST)
From: Joe Mazzarella 
To: Roc Cutri , Carol Lonsdale ,
    John Good , Tracey Evans 
Subject: Todays's DBWG meeting. See http://crystal:8000/2mass/

Since Roc was at the rescheduled project meeting, and Carol is out today,
I thought I'd summarize highlights of today's 1PM DBWG(?) meeting.

Attended: John, Mazz, Tracy

Tracy: -Will likely be occupied mainly with DBMAN Subsystem Design Spec
        through Dec after pipeline deliveries this week.
John:  -design and doc discussions with Mazz 
       -Infrastructure System Design Doc (ISDD) begun
       -Architecture Methodology/Philososphy Doc will be updated.
       -writing/testing "infrastructure" software and 
        experimenting with prototype Java apps.
Mazz:  -working on design docs and prototype interfaces
       -started new/updated 2MASS Web pages
        See http://crystal:8000/2mass/, especially
            http://crystal:8000/2mass/dba_design_main_fr.html
       -outline of System Components Doc (SCD) nearly complete
       -linked in Carol's Science Requirements Doc (SRD) using latex2html
       -Applications System Design Doc (ASDD) begun; 
        it will have links to SRD 

We discussed some basics of the WSDB: 
~300 characters per band times 3 bands ~ 900 chars per record
We estimated in rough terms the amount of disk space per column of data 
in the WSDB:
    0.5 TB = (5e11 bytes)/(~50 columns)
           = 1e10 = 10 GB per column!

We jokingly discussed a possible Sybase/IQ-type configuration with
effectively 1 column per 9GB disk which would minimize the amount of data
that has to be carried around (cf. long records) for query opimization.
More likely the flat pipeline WSDB will be split into 2-4 Informix tables
for query opimization. John will be experimenting with these issues. 

Some actions: 
John & Mazz: will meet Thur. for next stage of design and docs.
Roc: We still need a 'public' (NFS-mounted, not local) disk partition
which is routinely backed up for the development team to work 
on Web/DB science analysis code. In the mean time John and Mazz 
are working on our local disks.

*Discussion after meeting with Mazz: Roc will ask system group to
 turn lugosi:/d1 into lugosi:/2massdba, export it to
 the development team and put it on a backup schedule like /2massc.
 
Mazz: Will work with Roc to configure other 2MASS docs into the new Web.
      Will talk with Linda and Roc about QA-specific components of the
      Web-based analysis system and fold this into the evolving ASDD.

All:  Please comment on initial Web updates and send pointers to
      relevant stuff we should have linked in.