IBM Apptio

 View Only

 Partition Data by Date

Jump to  Best Answer
  • CostingStandard(CT-Foundation)
Apptio Community Member's profile image
Apptio Community Member posted Sat December 04, 2021 12:54 PM
Hello,

Partitioning data by date allows one file to be uploaded that has the date in each row so the data in the file is reflected in the appropriate period.

A file that has charge number and account budgets for each month can be upload and through data date partitioning the budget is loaded in the appropriate time period for each charge number and account.

I am successfully using this in cases where I have one file that I upload at the beginning of the year. and if there are changes to the file during the year overwriting the previously uploaded file.  This results in the entire model for all time periods being updated for the entire year. 

What I am unsure of is what the expected behavior is where one file is loaded as the initial upload and then rather than replacing it when there are changes another copy of the file is uploaded to a later time period like June or July?  From what I am seeing it appears to ignore the file uploaded in June or July.

Any insight you can provide regarding this would be greatly appreciated.

Thanks,

Gregory
#CostingStandard(CT-Foundation)
Alex Edwards's profile image
Alex Edwards  Best Answer

Hi Gregory,

Thank you for the question. 

Based on what you have described, the behaviour sounds as expected and isn't actually related to the fact you have used a Date Partition in the transformation.  

If your initial upload was in January and then you made just one ad-hoc upload in July, the month you currently have selected in Cost Transparency will determine which data set is being used. Rather than the initial upload being ignored, your ad-hoc in July upload simply doesn't apply until you are in the month it was introduced.

Using this example, when viewing CT from the month of March, this will draw from the initial upload you made in January. In fact, this will be the case up to and including June. But once in July, Apptio sees your ad-hoc upload as the most recently data source and will use that data set until there is another upload. 

Assuming you have your Data Refresh Cycle set at 'Ad-hoc' then you will be able to choose to make one monthly upload at any point you wish. For more information on this please see the 'Data Refresh Cycle' section of this page of the Help Center.

https://help.apptio.com/en-us/studio/data_studio/upload-data-file.htm?Highlight=table%20upload

I hope that answers your question, but please let me know if you would like to discuss further.

Many thanks,

Alex







#CostingStandard(CT-Foundation)
Guillermo Cuadrado's profile image
Guillermo Cuadrado
Hello, @Gregory Donaldson. I haven't tested this thoroughly, but it probably depends on 2 things:
  • Whether you have defined the dataset as Ad-Hoc loads
  • More importantly, on what period you load the updated versions of the file
We do something like what you mention for the Budget & Forecast in CT. The first load (in JAN) runs until we load the first Forecast (e.g. in APR).
I believe it works as I expect it to.​
#CostingStandard(CT-Foundation)
Apptio Community Member's profile image
Apptio Community Member
@Guillermo Cuadrado and @Alex Edwards

Thank you both for helping me get my head around this.

I did a much simpler test.  What was confusing me was that:

  • Values changed in the updated file that are for the present or prior periods to when the file is uploaded are not reflected in the present or prior periods and remain what they had been.  Updates are only applied going forward of the period of the upload.

With respect to the Data Refresh Cycles they don't affect the data uploaded but rather are used to check if the data in the model is current based on what was specified when the table was setup.

Thank you very much!

Gregory
​​​​
#CostingStandard(CT-Foundation)
Apptio Community Member's profile image
Apptio Community Member
Hi,

It's been a long time since I used this implementation, but it worked as @Alex Edwards described.

There was a potentially interesting Use Case which came out of it, based on us loading full year financials each month (Actuals + Forecast) and calculating them with Date Partitioning. By combining each month's output manually we could track how changing FY values affected any item over time, from the Unit Cost of an individual Service through to the total cost base.

I would recommend that you don't do that, since most people found the output so confusing that we gave up on it and just went back to showing the latest available data! 

Jon
#CostingStandard(CT-Foundation)