[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: 980428: NetCDF, Re: 960111: Unidata

Hi Francisco,

> >                                               ... Recently, however, I have
> > embarked on a project to improve our assimilation and interpretation of
> > real-time data from oceanographic moorings, satellites and ships. The first
> > effort is primarily focused on the mooring data (which also includes a small
> > amount of meteorological information, e.g. wind speed and direction). We are
> > trying to decide if we should use a relational database or something like
> > netCDF. As you know each has its pluses and minuses. Most (all?) of your
> > products are geared toward Unix/netCDF, correct? Will that continue to be 
> > the
> > case? We work on both Windows and UNIX (HP 10.2). Any advice, comments would
> > be appreciated.

Probably the most important factor in deciding how you store the data is
how the data will be accessed.  Something like netCDF for a data archive
may be appropriate if the data is written once or appended to existing
data in a structured way, is accessed as named arrays with associated
attributes, and is intended to be accessed by application programs
written in C, Fortran, C++, perl, or Java.  A relational database would
be more appropriate if the data was updated frequently, access to the
data was to individual values rather than arrays, access by key was
required rather than by index, and access in unanticipated ways by ad
hoc query was envisioned, for example through SQL queries.

Most of our applications have been targeted for Unix in the past, but we
are now developing primarily in Java with the goal of platform
independence.

--Russ

_____________________________________________________________________

Russ Rew                                         UCAR Unidata Program
address@hidden                     http://www.unidata.ucar.edu