Purpose of Clean House?

[cccschip]

I'm new to Transformer and trying to understand the purpose of the Clean House utility.

The cubes we have created so far are rebuilt each night.  It is my understanding that this rebuild fully refreshes the categories in each dimension. 

But, after reading the doc about Clean House, I am doubting this.  The utility makes it sound as though dimension categories are not deleted during the cube building process.  True?

That would make sense, though, because such a loss could cause reports to break.

Clarification appreciated!

[Grim]

Answer here: http://www-01.ibm.com/support/docview.wss?uid=swg21349059

[cccschip]

Funny thing...that's exactly the doc I read that made me believe our prior assumptions were wrong and led me to post this question.

So, rephrasing a bit differently...

When rebuilding a cube (without cleaning house), are categories that no longer exist in the source data removed?

[cognostechie]

That would make sense, though, because such a loss could cause reports to break.

Only if the Category Code property for the levels is left blank.  If you put something in that property, Transformer will assign the same category codes every time the cube is built regardless of Clean House
or anything else.

[bdbits]

Clean house will remove categories unused since the date you specify, as specified in that article. I do not believe old categories are otherwise ever automatically removed, hence the presence of clean house in case you do want to get rid of them. However in most cases, a truly 'unused' category is not going to be associated with any facts, and so it will not be visible to end users in the cube. (Maybe an MDX query could see it, I do not know.)

Categories are saved with the model when you save it in Transformer. If you rebuild from within Transformer, or regenerate the categories, any updates will be lost unless you re-save the model. If you rebuild from the command line and use the option to create categories (as is often done with nightly builds), it will do so for the cube build; but whether it saves them back to the model depends on other command-line options. So these various circumstances are going to affect whether you see old categories when you open a model file.

cognostechie raised a good point - if you override the generated Category Code then it will be retained with your value. We nearly always let category codes get generated and have not had problems over the last 3 years or so with anything breaking.

I don't know if I added anything there, but maybe different wording will clarify it for you.

[cccschip]

Thanks, all.  I especially appreciated the rephrasing by bdbits.  I hope to digest this better when I get more immersed into Transformer and cubes.

[cogcurious]

I was reading the topic for getting clarity on clean house. What do we mean by an unused category?

[RKMI]

Unused category are intersections which were created on a previously generated categories/ cube build but, these intersection might have been removed in the data source. To avoid reporting on removed category we use clean house operation to regenerate the most current valid intersections/ categories.

Thanks,
RK

[wyconian]

Hi

I use clean house a lot.  My client doesn't want to track any slow changing dimensiosn so when, for example, a new manager is assigned to a cost centre they only want to see the current manager and not see how that changed over time (personally I think that's a bad requirement but that's what they want, as a lowly contract I always do what my client wants).

This means when something changes I need to clean out the model and reload everything which is when I use clean house changing the date to tomorrow so everything gets cleaned out.