Line-Item Dimension

[#1]

Dear Experts,

I have a confuzion abt a Line-Item dimension. When you create a Line-Item dimension for a particular characteristics whose cardinality exeeds the limits significantly.

So it saves the creation of new Dimension table and so the Disk space. It reduces join of 3 very large tables(fact, dimension, SID) to a join of two tables(fact and SID).

My question is, if these many advantages are there after creating line-item dim. Then why dont we create every characteristics as Line-item dimn.

If above thing is not possible, then what are the restrictions on it? And what are the differences in Dimention table and SID table?

Points will be rewarded.

Thank you.

[gnanamk]

Hi Harshal

We depend on multi dimensional model to increase the overall performance.

It's main impace to analyse data in multi ways.

line item dimension can hold only one characteristic and so limeted to 13 ways only because of customised dimensions are 13.

so its almost near to star schema concept and no use of sid concept and the advangates of extended star schema concepts.

hope it helps u.

[Former Member]

Also a line item dimension can have only one characteristic and hence if you wish to club related characteristics together - line item dimensions cannot be used..

Arun

[Former Member]

However, it could also be argued that aggregates tend to increase performance by creating flat dimensions hence if you are using less than 13 characteristics why not create individual characteristic dimension and make them line item. Arguments against this way of modelling is that additions in the future are not being considered. However, my thoughts would be this is a could be analysis which when required can be considered and what is more appropriate for now

[Former Member]

Hi Harshal,

First of all we only have 13 dimensions to define by ourselves apart from the 3 mandatory system defined dimensions. <b>And in a line item dimension we can have only one characterstic</b> in that dimension.

So if we mark every dimension as a line item dimension we can only have 13 characterstics in a cube at the max. This results in many confinements in modelling, so it is advisable to have only highly granular characterstics in a line item dimension.

Hope this helps,

Regards

Message was edited by:

samara reddy

[Former Member]

Thanks a Lot.

[Former Member]

When you mark a dimension as line item it is included in fact table, because of that you avoid the join between big tables.

If you put al characteristics as line item you will have all cube data in fact table and you will lost every adventages of multidemensional modeling.