Skip to Content
Technical Articles

How to Optimize BPC Model in SAP BPC 11, version for SAP BW/4HANA

Cube-like aDSO is generated as basis provider (part provider) SAP BPC 11, version for SAP BW/4HANA (BPC 11) for each BPC model. Together with this change,  process chain /CPMB/BUILD_INDEX and /CPMB/FULL_OPTMIZE are now obsolete, as result, there is no out-of-box way to optimize BPC model (aDSO) either from BPC web admin or data manager.

By reading note 2669682, you may notice it’s possible to leverage the new BW process type ASDOREM in BPC 11 to optimize the model. In this blog, I will briefly describe you how to create a data manager package to achieve this. In the end, user could trigger this package no different to what he can do in older BPC release.

  1. Create a BW Process Chain for BPC Model

a) Goto tcode RSPC in SAPGui, and create a new Process Chain.

b) Create a step with process type ADSOREM

In process type navigation pane, find ‘Clean up Old Requests in DataStore Objects(advanced)’ (tech name ADSOREM)

             

i.  Create new variant for this step

ii. Input/select the tech name of the aDSO (of the BPC model) in first column ‘DSO(advanced)’

iii. Select ‘Activate Requests’ for column ‘Clean-up action’

iv. Input number of requests you want to left for compress( Select All but the X latest request) in column ‘Reqs’, optionally you can also input the days in next column as ‘Select all requests older than X days). In my example, I input 0 to clean-up(compress) all requests.

                 

 c) Create two steps with switch mode of aDSO

It’s only possible to run clean-up when aDSO in load mode, so we have to switch aDSO to load mode first and then switch it back to Plan mode after execution.

In process type navigation pane, find ‘Switch DataStore Object (advanced) to Load Mode’  and ‘Switch DataStore Object (advanced) to Plan Mode’, and create two steps with them. Please input the same aDSO tech name you used in step b).

d) Create a Start process

In process type navigation pane, find ‘Start Process’ and create a step with it.

 

e) Define dependency between steps

Now you should have four steps in the process chain, we need to define dependency with it.

We start with ‘Start’ step, right click it and select ‘Connect to’ -> ‘Switch to Load Mode’; then you link ‘Switch Load Mode’ to ‘Clean Up DSO Request’ then to ‘Switch to Plan Mode’ in the same manner.

Please note that you when link to ‘Switch to Plan Mode’ please make sure that you select the third option ‘always’ as you need to switch back to plan mode no matter the previous step succeeded or not.

When it’s done, you should have something similar like this in network view. (Too see network view, go menu Setting -> Network display)

f) Check and Activate

 

 

  1. Create a BPC Data Manager Package to trigger BW Process Chain

 Now it’s time to create a Data Manager Package in EPM plug-in to let end user to trigger this process chain when needed.

a) Log-in EPM plug-in to the BPC model and go to Data Manager Ribbon

b) Go to ‘Organize Package List’ and add a new package

c) Select process chain ‘/CPMB/TRIGGER_BW_CHAIN’, this chain is delivered by BPC to trigger any BW process chain.

d) Modify Script if you don’t want end user to select the BW process chain (created in previous step)

i.  Select ‘PROMPT’ on left and de-select the checkbox of column ‘Activate’. This is will hide the prompt for BW process chain name input.

ii. Select ‘TASK’ on the left and input the name of created BW process chain in step 1 into column ‘Value’.

iii. Save to leave this popup.

e) Give a name/description and save this new package.

       

  1. Execute Data Manager Package

Run this package from Data Manager, once status is showing ‘Succeeded’, we can go and check the backend.

Goto BW Modeling Tool (Eclipse), find aDSO of BPC Model and select ‘Manage aDSO’ from the toolbar.  There you can see that all the requests are activated.

 

NOTE:

  • You need to create a process chain for each model you want to compress.
  • Process type ADSOREM today can only compress records (per key combination), zero elimination is still NOT possible at the moment.
  • Above BW process chain is just a simple(basic) example, in your real use case, you can add other process types to fulfil your user requirement in one execution.
25 Comments
You must be Logged on to comment or reply to a post.
  • Do you think that in some SP standard chain LITE_OPTIMIZE will be corrected? And what about zero elimination? In some scenarios zero elimination is a required functionality….

    • Hi Vadim,

      Just a quick question to you around zero elimination. Can you name few scenario which require it? And usually how often it’s scheduled?

      Best regards, Willliam

      • There are a lot of cases where zero records are generated:

        1. Clear data using CLEAR DM package (example: clearing forecast/budget category to reuse it, clearing incorrect data load, etc…)
        2. Clearing data using script logic *REC(EXPRESSION=0) for some business logic calculations.
        3. Clearing data using script logic *RUNALLOCATION
        4. Clearing data as a result of rules execution
        5. Clearing data on data load with option to clear records

        After clearing data we still have a record with zero amount

        Result:

        1. Impossible to distinguish between no record (empty cell in report) and zero result – especially for parent members (sample – you are looking on some parent – if the result is empty then all dependents are also empty; if the result is zero then it can be +1000 -1000)
        2. Zero records will be processed in script logic – performance issues
        3. Queries will also process zero records – report performance.
        4. Impossible to delete dimension member if some data was posted on this member (in previous BPC versions you delete data and perform compression with zero elimination)
        5. Export data will contain useless zero records
        6. etc

        It’s a good idea to schedule compression with zero elimination over night. But it can be launched manually after some significant cleanup

         

  • Hi Vadim,

    No, Standard chain LITE_OPTIMIZE will NOT be back.

    Fully understand the requirement of zero elimination, but need to check with BW/4HANA team on this.

    Best regards, William

    • Please do William,

      I just had an SAP advisor tell me on a incident that ADSOREM (recommended by SAP in note 2669682) is not BPC relevant.  It may historically be BW namespace but when you recommend it as replacement for a BPC process your BPC guys need to know.  Also it doesn’t like large volumes of both records and requests and is effectively slower than the old Lite Optimize which I understood performed more tasks?

      Best Regards

      Peter

  • HI William,

     

    Re  “You need to create a process chain for each model you want to compress.”.  Not strictly true, I have a chain that runs for multiple models.  You can even use wildcards in the variants to select all BPC objects of selected type ie /CPMB/*.

     

    Best Regards

    Peter

    • Hi Peter,

      You are not wrong. But here, we are triggering the compress from data manager usually is model specific. Of course you can create a process chain processing multiple aDSOs, but it’s not suitable to be triggered from DM conceptionally.

      Best regards, William

  • Hi William Yu

    Is it necessary to compress BPC aDSOs for performance purposes, like we used to with light optimizes? We have followed the guidance you set out in your blog and created process chains to clean up old requests in aDSOs, but we are finding that the process can be very slow due to the way that aDSOs manage requests. Per Gregor Dieckmann in a question posted to the community here, a new request is opened on the aDSO each time data is saved. This means that we get a high number of requests for the number of records, which is then much slower to compress.

    For example, over the course of a day, our users generated ~420,000 records, which were captured in ~20,000 requests, i.e. ~21 records per request. As a result, the compress took 83 minutes.

    Thanks,

    Hugh

    • Hi Hugh,

      As a non-expert on HANA, I assume it might be case by case.  I would expect at least compressing aDSO will release some HANA memory(not sure how much) especially when doing a lot of simulation and iteration.   Maybe you can also obverse from your side around performance and memory.

      On the other hand, I’m not sure if the time spent on compression is normal. Maybe it worth a further check with BW team. If the ROI of aDSO compression is low, probably you only need to do it once in a while.

      Best regards, William

  • Hi William,

    My system is SAP BPC 11.1 NW, Standard Model.

    I need to delete the zeros because I need to remove the master data.

    We have a problem with a machine, and I need to create a clean environment of master data and transactional data.

    ¿Any suggestion to delete zeros?

    Best regards, Alejandro.

  • Hi William,

    We have just upgraded to BPC 11.0 & BW/4HANA 1.0 SP15 last week. Now we did not get an answer from SAP on any alternate solution for performing zero elimination, the suggestion was to upgrade to BW/4HANA 2.0 SP04. What is your recommendation for zero elimination? Selective deletion from the ADSO?

    Looking forward to your response!

    Regards,

    Swakshar Choudhury

  • Hi Swaklshar,

    I’m afraid that this feature won’t be down-ported to BW/4HANA 1.0 at the moment. Probably you need to check with BW expert on viable options. I assume there is no solution from BPC side.

    Best regards, William

     

  • Hi William,

    we are on BW2.0 SP04, and I tried to perform a zero compression on last request of a BPC 11 model (supposedly collecting backwards already activated requests).

    BPC4HANA 200 SP0003 / DW4CORE 200 SP0004

    The log indicates that Zero Elimination has been performed but the BPC Model still holds zero values…

    Correction: after a new test generating zero balance on full key, it did compress and delete the zero value from the last new requests.

    But, still remains the zero values from previously activated requests (without having flagged them with zero elimination) they don´t seem to be captured and eliminated on last request activation.

    Oss 2669682 Version 9 updated yesterday (01.09.2020) still states the same as previously: not supported…

    Hopefully this will be enhanced soon, since I found two dumps related to in ST22 (ASSERTION_FAILED CL_BW4_MANAGE_PROCESS=========CP)

     

    Best Regards,

    Luc

     

  • According to OSS 2669682 Version 9 dd 01.09.2020 Zero Elimination it is not supported in BPC11. This must interpreted in a sense that Model Optimization cannot yet be set up with Data Manager Package as described in present blog. And hopefully, it will in the future.

    “For data mart Data Store objects, you can now select zero elimination when activating requests.
    When zero elimination is applied, all records in the active table for which the total for every key figure is zero are deleted.”

    Cfr.

    DataMart ADSO for Planning Zero Elimination

    This means that it can be manually done for each BPC Model thru BW/4 Manage Data Store Cockpit with or without zero elimination.

    First, switch the BPC Model Adso into Load Mode.

    Please beware that if requests are activated without zero elimination, because one forgot to flag accordingly it won´t be possible to clean them up afterwards in one go: then it will be required first to insert non-zero values on the key still holding zero values, then in a second step clear them or zero update is done, finally activating the last request with flag zero elimination. Then switch back the BPC Model Adso into Planning Mode.

     

    Best Regards,

    Luc