topic Avoid Or Keep Synthetic Key(s) in QlikView

Avoid Or Keep Synthetic Key(s)

Anonymous — Fri, 31 May 2013 20:29:01 GMT

I'm working on a data-model that is having Synthetic Keys:

DReader:

ReadID,

ContactID,

Other Fields

CReader:
ReadID,

ContactID,

Other Fields

Pub:

ReadID,

Other Fields

Contact:

ContactID,

Other Fields

Should I keep the Synthetic Keys or is there a better way to approach this?

Re: Avoid Or Keep Synthetic Key(s)

vivientexier — Fri, 31 May 2013 20:51:31 GMT

It is better to avoid synthetic keys. If you let QlikView create this kind of key it will take more time to reload your application. You can test if you have a large amount of data you will notice a significant difference.

Moreover, data model is hard to read if you have to many synthetic keys.

Re: Avoid Or Keep Synthetic Key(s)

Miguel_Angel_Baeyens — Fri, 31 May 2013 21:03:48 GMT

Hi,

Synthetic keys are not bad by themselves, meaning when done or left on purpose, which in my experience is very unsual. As QlikView links tables when one or more fields have the same name, it can lead to mistake or do worng associations between tables if fields are not properly renamed (e.g.: Id in invoice and Id in customer, same name, different meaning).

I'd rather do the composite (synthetic) keys by myself, and adding some autonumber() function so they use less memory and less space into disk, when stored.

Hope that makes sense,

Miguel