Archival of Data

[Steve Vincent]

You know when someone makes a seemingly innocent remark and then your TM1 brain starts peddling in to overdrive about the reprocussions of those words? I've had one of those this week and it's depressed me as it means a nasty, horrible change to the models to deal with it

I'm dealing with data that (normally) is only collated and uploaded twice a year (although it takes several months to do so). Each time it's loaded it goes to an element in the "version" dimension. Due to the complexities of the hierarchies and the way they relate to the business, if a department is moved from area A to area B after the 1st version has been agreed, it will change the consolidated values in some cubes but not others. This is because some cubes have N levels that equate to consolidated levels in others, and if the move affects a big bunch of numbers then it's going to bugger up the historic picture and put cubes out of sync.

So i'm left with (what i think is) the only option of creating a set of archival processes to create copies of the dimensions, attributes, cubes and then copy the data across. I know that's something done a lot of the time, but the model i inherited wasn't designed to work in that manner. Normally i'd just be manually recreating all this and copying data with TI, but the archival is to be done by people who don't know how to create models and nor should they need to.

Sooooo, my question is this. has anyone already written a suite of TIs to do such work? I have started planning out what i'd need to do, but if anyone has already done this kind of thing it would be fantastic to have a look at how you went about it If not, then i'll try to write this suite in a gereric way to allow others to benefit from my own pain

[Michel Zijlema]

Hi Steve,

Not a direct answer to your question, but another possible solution to (slowly) changing dimension structures could be to split the dimension into more than one: a dimension with all (lowest level) departments and a dimension with the department groups. This way you can load data on different combinations of departments and department groups. The rule to the cube on a consolidated level would also be easier: directly link on department groups and take department total for the departments dimension.
Of course you would need to rely on zero suppression if you want to have a hierarchy view on the deparments.


Michel

[Steve Vincent]

interesting idea...unfortuantely i can't see the users going for it (too hard to navigate) plus it gives the (substantial) possibility of a user loading data to the wrong "consolidation". Half the time they don't know where in the org they sit anyway, so without the hierarchy in the dim itself i can see it causing more problems than it'd solve. One of these cubes already has 12 dims too it and i don't like the number 13...

My archival Tis are going OK so far, we'll see what comes out of it.

[pobrouwers]

Hi Steve,

I'm very interrested in you Archival TI's.
Do you finally achieve your work ? Can you share it with us?

Thx

Pierre

[Steve Vincent]

http://forums.olapforums.com/viewtopic. ... 8230#p8230

Sarah has also been looking in to it. I've not had time to put my processes in to that thread yet but i'll try to before the end of the month.