CLean house

[cognos05]

Hi,

I have cube with days dimension from 1 to 31 and data loads for customer on a daily sales. Now when the next month 1 comes the data is going to be appended to the previous 1 sales value. I just wanted only that days data to be loaded.

So do I need to do a Manual clean house on 31 st , write now we have a script to build the cube nightly

cogtr.exe -n2 -c -g -m

Please advise.

Thanks,
Nithya

[Michael75]

Normally, according to the info you have given, your cube is going to be rebuilt from scratch each night (think TRUNCATE), so you don't need to worry about appending data.

Unless, that is, you have checked the option "This cube is incrementally updated" on the Processing tab of the cube Property sheet. And if you have done so, do you really need this?

Assuming that you are incrementally updating, and from the info you have given, your days dimension is keyed on month and day, but not on year(?). This might not be an optimal design.

In my early days with Transformer, I ran Clean House a couple of times. Mostly I got away unscathed, but one time I lost an alternative drill-down (no backup of model) and had to recreate it manually. I haven't tried it since, not seeing its usefulness. I'm sure others here with far more Transformer experience than me can weigh in here...

HTH
Michael

[bdbits]

I do not recommend clean house either, based on bad experiences trying to use it. And I don't think it will solve your problem anyway, as it relates to dimension/category generation.

What did you mean the data will be "appended"? Is your input from a database? If the cube is rebuilt each time (likely), it should only reflect the data in your source at the time you run it.

If you are doing it incrementally and want to "start over" once a month, you could add some steps to delete the target cube file(s) before building it.

[cognos05]

Always the file that loads cognos transformer gives data from1 to 31 and no months, just the days.

Say on Jan 1 to 31 the cube is loaded with values and on feb 1 the file will have day1 and not feb 1 , so say I have a customer X whose sale value on 1 jan was 500, and now on feb 1 is 300 .
the cube will show 800 since it know history .

I want something like - cube is going to be rebuilt from scratch each night .

To achieve this which property do i need to set.

Thanks,
Nithya

[Michael75]

To achieve this which property do i need to set.

The answer to your question is in the reply I already sent you.

Always the file that loads cognos transformer gives data from1 to 31 and no months, just the days.

But you wrote in your initial post:

Now when the next month 1 comes the data is going to be appended to the previous...

?!?! Does your dimension distinguish between months or does it not??? The clarity of the answers you receive depends on the clarity of the info that you give.

[cognos05]

I have a Customer dimension say customer 1000,2000. Products dimension like ABC.
Days - values from 1 to 31 no months involved. Just numbers.

SO example jan3 - data comes as Day -3 , Customer -2000 ,Product -ABC, amount -500

Now Feb 3 -   data comes as Day -3 ,Customer -2000 , Product -ABC ,amount -100

so on feb 4 I should see only customer 2000 with value 100 on product ABC for Day -3

So should I uncheck the incremental update ? what does it do .
so rebuild from scratch everyday what property should I set ?

Please advise
Thanks,

[bdbits]

Yes, incremental adds data to what is already there. For what you described in that last post, you do NOT want incremental.