Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!utgpu!water!watnot!watmath!clyde!cbatt!gatech!seismo!tiberio From: tiberio@seismo.UUCP Newsgroups: comp.databases Subject: Normalized Databases Message-ID: <43239@beno.seismo.CSS.GOV> Date: Wed, 1-Apr-87 08:46:19 EST Article-I.D.: beno.43239 Posted: Wed Apr 1 08:46:19 1987 Date-Received: Sat, 4-Apr-87 06:00:34 EST Distribution: na Organization: Center for Seismic Studies, Arlington, VA Lines: 35 Keywords: normal form, relational At our facility we have a certain large investment in a relational database design that is not in the highest normal form. We have some folks proposing that we redesign our database (and no doubt out database interface software) to a greater or highest normal form. They state we will enjoy the following qualities when this is done. 1) The ammount of space required to store the data is lowest. 2) The ability to update the database is greatest. 3) Explaining the database is simplest. While I agree with the first point, If disc space is not a problem I don't see where this is a big win. I'm not really sure under point two if they mean update the database or update the database design. We update (or replace in INGRES terms) our databases very little once they are established. I disagree with the third point, especially if I have to explain this twisted mess of relations (56 vs 24) to our mythical "joe" user. What I am asking you folks in netland is for some guidance. I like the intellectual concept of normal form relational databases, but I need to know if there are any good reasons not to go that route. I would also like to know if there are other advantages that the proposers failed to mention. I might add that we seem to have plenty of money for hardware, but that man hours are at a premium. Since this group has very little traffic, I don't see why we can't discuss this on the net, but please feel free to send me E-mail for any mega-responses. References to texts and/or papers would be welcome. beno!tiberio seismo!tiberio