>4.  What is the importance of client-server, distributed and relations
>database technologies to collection management databases?
 
        Actually, I would like a crack at all the questions that Kathy
has posted, but thought that I would exercise some self constraint and
only answer the one that I thought most interesting.  If Kathy or anyone
else wants to publish this or anything else, that fine.  This is
intellectual freeware.  Just as long as credit is given where due.
 
        As for the question.  I find the phrasing to be interesting.
I would argue that the only important thing that a database must be
nowadays is relational.  A database need not having any client/server
or distributed functionalities because these functionalities are
being dealt with at a different level than the database itself.  Let me
explain.  Lots of people are working on Internet retrieval tools and the
such that incorporate databases.  So, for example, the World Wide Web
now has an interface that allows anyone with the proper client to
conduct searches on an SQL compliant database.  That is, SQL databases
have been subsumed by the Web.  Same thing goes with the distributed
database functionality.  Databases will be linked with each other
through tools outside the database itself.  Already, the search mechanisms
in gopher and WWW allow multiple indexing.  Choosing a database on
either a client/server or "distributed" basis is not a good idea.   These
technologies are not database specific.
 
        However, it is still importanto have a database that is easy to
manipulate and that has a good framework.  Even if the manner of
"presentation" of the database does not matter, the database information
content is extremeley important.  A relational database is a way to
maximize information content and felxibility while minimizing the
amount of space used.  Basically, instead of storing all your information
in one big flat file, you store chunks of related information in separate
databases that can be linked together.  This "relational" model makes
a lot of sense.  You dont throw all your papers in one big folder.  You
organize them into small folders with cross referencing to other folders.
Relational databasing is not only organizationally wise, but also allows
for faster searching of information.
 
        The last point, one that I kind of almost make, but never
explicitly say is: Structuring and organizing information in a database
is fundamentally different from how the information in the database is
used.  That is, the way that information is extracted from the database
can be numerous, and can far removed from the database itself.  The
strcuture of the database is integral and the most important thing.
 
Cheers,
 
Robert Guralnick | Museum of Paleontology | University of California
Berkeley, CA 94720 | [log in to unmask] | (510) 642-9696