You are on page 1of 2

>>May be you have heard about normalization.

In this topic, we are going to take a look at what normalization is and how it came
into
being.
>>Okay, at the start off database design there was someone called Edgar F. Codd or
normally
referred to as Ted Codd and he developed the relational model for databases and as
part
of that he had a process called normalization.
And it very much is a process that we are going to run through, there is a number
of
steps that we would follow to normalize a database.
There is another term called entity relationship modeling and with that we are
talking about
entities.
Now an entity can be anything.
It could be a person, so it could be an employee or it could be a customer, but
also it could
be a conceptual entity as well.
So it might be an order or something that is not physical but we can still describe
it.
You know, you have adjectives about it and the properties of that entity.
When we go through normalization, we tend to end up with all of our entities as
individual
objects.
So we might have an employee or we might have an order or we might have a customer.
The process of normalization is very much applying rules to end up with that entity
relationship model.
If we just said where are our entities, then we haven’t really gone through a
strict
rule-based system and it is very very easy to get it wrong.
So normalization is a very strict and rigid approach and it ensures that we have a
well
designed database for most of our systems.
We are not always going to use it, but most systems will start off with
normalization.
So we are looking at our data and we are applying rules to it and using those rules
we can end
up with a well-designed database.
>>So Geoff, would I load my data, before or after I normalize?
>>That’s a really good question because many many many years ago when I started
working
with computers, I was working with databases and I did what a lot of people
typically do
and that is load an awful lot of data in, then subsequently find out that is not
well
designed and then I need to change that data and change the structure of the tables
to
make a well designed database.
The problem is with no data, it is impossible or near impossible to normalize
something.
We do not know what the structure could be because there is no data there.
With too much data, then we have a situation where we have got to fix it, we have
got to
move data around, change structures so I would say we would start with a sample of
our data,
something that is representative of what we are going to store so that we can
understand
how the entities within there relate.
But if we put too much data in, then we have too much work load to get it back to
something
that is well designed.
So that is an overview of the history and the concept of normalization.

You might also like