Comments on: Parent-Child Dimensions in Analysis Services – Performance Walkthrough

By: Dirk

Dirk — Sat, 31 Oct 2015 09:33:49 +0000

Thank you for your Post! My question is: How could Rollup (+, -, ~ would be required) be implemented?

Yours
Dirk

By: Paul Hernandez

Paul Hernandez — Thu, 05 Mar 2015 14:21:12 +0000

Hi,

great article.
BTW, I would like to ask again about dynamic security. If I want to give access to a user A to Fruit, a user B only to Red and a user C only to Lemon, how could I implement that using the bridge table technique. Usually you create a bridge table with the user key and the dimension attribute key you want to allow. Then you create a factless fact table using this bridge table and then create a Count measure. Then for the current user you proof if this count is empty or not in a given dimension. I cannot see how can I add members from other levels in this bridge table.
Any comment will be appreciated.
Kind Regards,
Paul

By: Nagesh

Nagesh — Wed, 11 Dec 2013 19:22:39 +0000

I am seeing some issues with the cube when I browse the measures with filter dimension hierarchy filters at multiple level (say DimTime, Year and Semester), whereas there is no issues (slowness) if I browse at same level like (Semister1, Semester 4) etc. I tried adding the aggregations of the attributes which are used in hierarchy. Any recommendations are highly appreciated.

By: xlcubedblog

xlcubedblog — Mon, 20 May 2013 10:59:58 +0000

Hi,

Not sure what you are asking – it’s still possible to control security in the same way as with a normal parent/child dimension. Is there a particular problem you are having?

Thanks

By: hxy0135

hxy0135 — Sun, 19 May 2013 18:26:28 +0000

Hi,
If you need implementing dynamic dimension data security on different level, say, some users can see data at level1, some users can see some members of level2 and/or some members at level3, how would you implement ‘Allowed member set’ in a user role editor?

By: xlcubedblog

xlcubedblog — Thu, 03 Jan 2013 14:52:15 +0000

Hi Bryan,

There is an option on the measure group “Ignore Unrelated Dimensions”, setting this to False will not display the parent’s data and the children’s level.

The other approach is to create a dummy child to hold the parent’s data.

By: Bryan

Bryan — Thu, 03 Jan 2013 14:39:14 +0000

Great post! I have one question; Creating the hierarchy this way, won’t you encounter errors if you loaded fact data for an intermediate level? Say your fact data was loaded for Level2 Red: if you select level1 as the measures used dimension, all of the children would show the data for the parent, which isn’t accurate, but if you showed the leaf level of the hierarchy, the cube would error on processing because no fact data is tied to it. How do you specify a level of granularity that will allow for differentiating the parent/child data without throwing an error at process time?

By: xlcubedblog

xlcubedblog — Fri, 21 Dec 2012 15:58:32 +0000

Hi Rudo,

In that case you could create a “Dummy” member at the lowest level and just load the data against that. You could give it the description: (Parent Caption) to differentiate it.
The other approach is to use a different measure group to store the values at the parent levels and then bring them together in one calculation that combines the measures from each group.

Hope that helps!

By: Rudo

Rudo — Fri, 21 Dec 2012 13:56:51 +0000

Really nice article ! Just one question, what in the case when I have parent-child hierarchy where data are also at the non leaf level of the hierarchy?

By: xlcubedblog

xlcubedblog — Thu, 22 Nov 2012 08:40:52 +0000

No, since the lower IDs are different, Analysis Services knows that it’s not a duplicate, just that the parent member is the same.