DESTINATION_APP vs LOOKUP for data movement betweeen models

[Former Member]

Dear experts,

I would like to ask about the difference and Costs/Benefits of using each of this approaches to move data between models of the same environment at an aggregated level. For example, obtaining a SUM of all accounts (parent member?) from INCOME model into a dummy account on EXPENSE model to use as a allocation driver.

As far as I know, both approaches could achive what I want, But would be helpful to know from your experience wich one is best, flexibility and performance wise.

To use DESTINATION_APP, script logic should be on the source model to later be called on destination model using a RUNLOGIC_PH statement, while LOOKUP would only be necessary to implement on target model.

Thank you all for your time!

Best regards,

Joaquin.

[former_member186338 · Accepted solution]

Hi Joaquin,

In most cases "push" approach by DESTINATION_APP is better, only existing records will be transferred to target.

With LOOKUP and empty target you have to use very slow *WHEN_REF_DATA = MASTER_DATA (affects all dimensions).

Vadim

P.S. And yes, RUNLOGIC_PH is fine for: clear destination, push data from source, additional calculations in target...

[Former Member · Accepted solution]

Hi Vadim, thanks for your answer. I have two more questions regarding this issue and to close this thread.

1. Using DESTINATION_APP and RUNLOGIC_PH to move data between models, is it posible to have more than one source model to write on a single target model on the same ACCOUNT/AUDITTRAIL and accumulate values from the different sources. Somewhat like WHEN/ENDWHEN works with different RECs in the same scope. As I see it, using this approach forces you to have one logic for each source model, each one with different scopes, and one calling script on target model, making imposible to accumulate values on same target dimensions.

2. Whats the difference, performance wise, in scoping dimension on a XDIM_MEMBERSET statement and to scope them on WHEN/ENDWHEN? Which one is better?

Case 1:

*XDIM_MEMBERSET ACCOUNT = PL100

*WHEN ACCOUNT
*IS *

*REC(...)

*ENDWHEN

Case 2:

[...]

*WHEN ACCOUNT
*IS PL100

*REC(...)

*ENDWHEN

I've seen both approaches and still can't figure out whats the benefit of each. Is it better to have the most dimensions as possible scoped on XDIM_MEMBERSET or to have less?

Thanks again!

Best regards,

Joaquin.

[former_member186338 · Accepted solution]

Hi Joaquin,

1. No, it's not possible to accumulate data in a single intersection from 2 models in the third model! You will have 2 WHEN/ENDWHEN loops and the second will overwrite the first! But I can hardly imagine the business case for this scenario...

2. "Is it better to have the most dimensions as possible scoped on XDIM_MEMBERSET" - Yes. But if you write default.lgf - you have to use *IS ... in most cases.

Vadim

P.S. There are no universal solutions, sometimes you will have to test and sometimes - to write a badi...

[Former Member · Accepted solution]

Hi Vadim,

Business Logic is pretty simple. I have three models to budget different INCOME accounts, each one with it's own set of dimensions and logic, and one EXPENSE model. In the expense model we need to allocate certin values using the income by office as a driver. For this I was thinking to move data from the three INCOME source models into the EXPENSE model at an aggregated level (ACCOUNTS = NA ; AUDITTRAIL = COPY_INCOME ; by OFFICE) to later use RUNALLOCATION using this values as a driver, this way I'll be passing just the totals account needed for the allocation and therefore not duplicating the complete budget both in sources and target.

I think I'll be needing different AUDITTRAILs in target, one for each source model.

Best regards,

Joaquin.

[former_member186338 · Accepted solution]

"I have three models to budget different INCOME accounts, each one with it's own set of dimensions and logic" - different sales types? In this case we use different revenue accounts! Talk to finance people...

Example - we have advertising sales model and copy sales model - results are going to different revenue accounts in a single P&L model!

Vadim

[Former Member · Accepted solution]

Vadim,

Yes, each INCOME model has a different set of accounts to budget but we need to sum the TOTALs of each, by OFFICE, to use as an allocation driver in EXPENSE model, as I don't want to copy and duplicate, if possible, the whole budget and it's complete detail (which is not needed as I just need TOTALS by OFFICE) into the EXPSENSE model. Later we'll be moving data, both from INCOME and EXPENSE models into a P&L model by account for reporting and some calculations (example: taxes) on the complete P&L.

Best regards,

Joaquin.

[former_member186338 · Accepted solution]

Move data from INCOME to P&L by account - you will have to do it anyway. Then, using P&L model - calculate revenue driver and save it to EXPENSE model (script in P&L model with DESTINATION_APP to EXPENSE). Calculate allocations in the EXPENSE model. Move data from EXPENSE model to P&L model including calculated allocations.

We are doing the same for overhead departments expenses!

Vadim

[Former Member · Accepted solution]

Vadim, thanks for all your input. It's much appreciated!

Best regards,

Joaquin.