Optimal data model?

[btrompetter]

Hello,

need QlikView a special data model for the source?

Is it better to use a relational model or a multidimensional model?

Regards

Bastian Trompetter

[Oleg_Troyansky]

Bastian,

QlikVIew doesn't require any special data model, it can read data from one or more sources, organized practically in any way. I would not recommend messing with multidimensional models, though, unless you have no other choice. Relational data sources work very well.

Oleg

[btrompetter]

Hey Oleg,

thanks for the info. If we use a multidimensional model or a realtional Model with agregated data.Would QlikView support the pre aggregation, or is it better not to use pre agregated data?

Bastian

[Oleg_Troyansky]

Bastian,

QlikView will take just about anything, but it's up to you to determine what data is good to be used in your particulate case. As in many other cases, the most common answer is "it depends".

If your pre-aggregated data is at the level of detail that you need for the specific application - then go ahead and use it. If you expect to go deeper in the level of detail, than it's better to read detailed data and not mess with pre-aggregation.

If you are relatively new to QlikView, I'd highly recommend engaging an experienced consultant for your first project, to "get it right" for the first time.

regards,

Oleg

[Anonymous]

Bastian,

Regarding the pre-aggregation, here is a basic rule.
If you pre-aggregate your data, the application size will be smaller, so the performance should be better in most cases. On the other hand, whatever you pre-aggregate in the script is not possible to "un-aggregate" on the front end, so the lowest level available on the front end is the logical limit for pre-aggregation.

Regards,
Michael