Such data migration is doable with the host language interface to Sybase being
the most viable approach. Keep in mind, however, that one of the purposes
behind the original development of CDF, which applies to some extent to netCDF,
was to provide a vehicle for access and management of scientific data for
which the relational model was inappropriate. For example, in applications
at GSFC, Oracle has been used for the management of temporal, characteristic
and inventory metadata, while the actual data are either in a raw (i.e.,
original) archive format or CDF. If the former, then translation tools for
subsetting to CDF are provided. Integration is provided at the user interface
level, above that of the data managers.
Many years ago, we did experimentation with doing graphics from (meta)data in
Oracle and conversion from data in Oracle to CDF, both through the HLI.
The results were useful, but not efficient. In both cases, only point data
were involved, so the inefficiences for a RDBMS for multidimensional grids
was not relevant. Given newer hardware and RDBMS implementations, the results
should be practical today, but probably only for point data not aggregate
access to random blocks of multidimensional parameters.
I don't know what the structure of the data that you have in Sybase might be,
but the typical reasons data are put in Sybase or netCDF are fairly orthogonal
to each other. I would be curious as to what you specifically might have in
mind.
Lloyd Treinish
------------------------------- Referenced Note ---------------------------
Received: from unidata.ucar.edu by watson.ibm.com (IBM VM SMTP V2R2) with TCP;
Mon, 10 Aug 1992 20:45:32 EDT
Received: by unidata.ucar.edu id AA09898
(5.65c/IDA-1.4.4 for netcdfgroup-send); Mon, 10 Aug 1992 17:39:05 -0600
Received: from nettlerash.Berkeley.EDU by unidata.ucar.edu with SMTP id AA09894
(5.65c/IDA-1.4.4 for <netcdfgroup@xxxxxxxxxxxxxxxx>); Mon, 10 Aug 1992
17:39:03 -0600
Received: from hp850.mbari.org by nettlerash.berkeley.edu (5.64.1/1.34.6)
id AA02506; Mon, 10 Aug 1992 16:38:34 -0700
Message-Id: <9208102338.AA02506@xxxxxxxxxxxxxxxxxxxxxxx>
Received: by hp850.mbari.org
(15.11/15.6) id AA04411; Mon, 10 Aug 1992 16:38:54 pdt
Cc: scri@xxxxxxxxxxxxxxx (Rich Schramm)
Mailer: Elm [revision: 66.25]
Well asking for help building netCDF on
hpux worked so well, now for the fun stuff...
Has anyone out there had experience creating
netCDF files from a database product (such as
SYBASE) which responds to SQL. Im thinking
about how I might make use of netCDF in querying
our oceanographic database to get data into a
variety of graphics packages. (ie pplus,
matlab, ncar?, pvwave?).
Rich Schramm
