Re: Units of Measurement in the Database Model
Date: 12 Apr 2002 11:30:27 GMT
Message-ID: <a96ggj$l07$1_at_news.tue.nl>
In comp.software-eng John R. Lewis <johnrlewis_at_hotmail.com> wrote:
> I am designing the Data Model for a large enterprise application, and
> have run into a problem. I solicit your feedback.
> The Problem:
> How should the database be organized in regards to measurements
> recorded, and the units of measurement (UOM) used to record them?
> When recording large numbers of measurements, it is important to know
> what is being measured. A number is meaningless without its UOM.
In addition, according to [Fowler], a number can be meaningless:
- when you don't know how was measured; - when you don't know when it was valid; - when you don't know when it was known that it was valid(For a detailed look at time aspects of databases, see [Snodgrass]).
How to organise the database is depending on how you're using it. If it's just persistent storage for your object system, and all processing is done by the object system, not the database, you can get away with the 4th solution. If the database is doing the processing, you'll need to denormalize a lot.
Stephan Eggermont
Sensus, systems that make sense
[Fowler] Martin Fowler (1997) Analysis Patterns, Reusable Object Models [Snodgrass] Richard T. Snodgrass (2000) Deveoping Time-Oriented Database Applications in SQL Received on Fri Apr 12 2002 - 13:30:27 CEST