COMPUTER_BASE
1=AAA
2=BBB

SOFTWARE_BASE
1=Software A
2=Software B

COMPUTER_SOFTWARE_XREF
1=1
1=2
1=2

Quote:

---

I find this way of doing it a little bit Over loaded in informations... isnt it???

---

Yes and no

There is a technique in analysing data called "normalising". The technique is ultimately used to reduce duplicated data to a single instance without losing any of the implied information like relationships.

Normalising the data will help to ensure that there is less likelihood of errors occurring, that problem solving is much more simple and that changes to data (maintenance) is easier. You will also reduce the amount of memory required to manage the tables and the amount of disk required to store the data.

However, normalising also introduces overheads. You may need more tables, you may need more CPU cycles to reconstruct the data. The additinal indexes make the database more complex and requires more effort to maintain and repair when there is an error. You are more likely to need database transaction support to ensure data integrity.

The trick is therefore to know when to stop normalising.

In your case it depends how much data you are going to generate and what you are going to do with it.

I think that storing the PC name and application name in the table as you have suggested is absolutely fine. I'd only look at his again if you have more than 10,000 computers to index or you calculate that you will have trouble with disk space.

Quote:

---

hmm...
really? can't get the logic of the last one...

---

Agreed, there is an error in the table. See my last post on normalising "adding complexity"

The table should look like this:
Code:

---

COMPUTER_SOFTWARE_XREF
1=1
1=2
2=2

---