Re: normalisation vs denormalisation

True, there is valid arguments for both sides, but in my mind that is not the real issue. The real issue is that poor database design effects not only speed, but also maintenence of code, efficency of code, and storage of data. The way to design a good database, is to design it first in a normalized fassion. Of course, over the same set of data, you can create more then one database schema. Because of this, when you are designing your database, you need know what structure and organization works well for the task you are going to accomplish. Then, after you have finished this, if some data works better being de-normalized, then de-normalize it. You just need to be willing to pay the cost.

Of course doing the above requires a willingness of your oranaziation to allow time for database design. Most companies want to rush straight into developement and thus the database schema is one large un-normalized mess which is hard to maintain, hard to fix, slow, and doesn't meet the custormers needs.

Eric Mortensen
mortense_at_cibadiag.com Received on Mon Aug 12 1996 - 00:00:00 CEST