Data Model best Pratice

[brindlogcool]

In the data model which is the best practice combining the Multiple Fact Tables in to one single big fact table and Join the Dimension tables in to one table

or Keeping the multiple small fact tables and not joining the DImesnion tables.Considering there are chances of the One to many relationsip in the dimension tables and columns will be duplicated will increase in the size of the data model.One would argue Qlikview will consider only the identical values while compressing the data but when the dashboard is accessed all the values has to dispayed which will increase the size.

Which one is the best at the user performance.Consider if i am going to use the App as the offline app then file size very much important as the QVW file will be downloaded into Ipad.

[hic]

My experience is that you should concatenate the multiple fact tables into one, but keeping the dimension tables as they are. Putting all fact tables into one common fact table may however mean that you need to implement generic keys, (See more in http://community.qlik.com/blogs/qlikviewdesignblog/2012/10/26/mixed-granularity) which could mean that you need to change the dimension tables slightly.

But you should not join the dimension tables.

HIC

[stevedark]

Hi there,

You may want to take a look at this blog post where I give my thoughts on the various options:

http://www.quickintelligence.co.uk/perfect-your-qlikview-data-model/

Regards,

Steve

[#4]

Here is document, which may give you some ideas for data model:

http://community.qlik.com/docs/DOC-3451

[brindlogcool]

Thanks for the response.I agree with all your comments and definitely consider all the blogs and make use of it.Since i am developing this app for offline mode. I have to reduce the file size as much as possible by eliminating the duplicate rows ,null values and removing unwanted fields(because of joining the tables or any other means) I can create a single fact table but in my scenario it is not the granularity of the fact table. Each fact table is having different dimensions except one dimension which can be linked with all other.

[#6]

You create reporting data model.

So, less important is, how data is related in the source. Much more important is, how it should be assiciated to support your reporting needs!

regards

Darek