Planning Analytics

Planning Analytics

Get AI-infused integrated business planning

 View Only
  • 1.  PAW - Automatic feeders

    Posted 04/29/22 07:45 AM

    Hi all!

    Feeders can be the most mind-breaking and complex work for a TM1 modeler.
     
    In the Performance Modeler, there is the option of generating feeder reports which help.  It seems though that those reports do not consider feeders for dimensions with multiple hierarchies. 

    I would love to have some help or suggestions directly in the PAW modeling workbench. 
    The feeder generation is confusing when linking data between cubes with DB(). The fetched data then must be fed exactly reversed from the source cube unless there is some reliable cell data in the target cube to trigger the feeding.

    The option of automatic feeder generation or feeder suggestion would be very nice when working in the PAW cube-rules editor. 
    Is this something for future consideration?

    Thanks, Asgeir



    ------------------------------
    Asgeir Thorgeirsson
    ------------------------------

    #PlanningAnalyticswithWatson


  • 2.  RE: PAW - Automatic feeders

    Posted 03/10/23 05:25 AM

    I would love to get some comments on this ...IBM?



    ------------------------------
    Asgeir Thorgeirsson
    ------------------------------



  • 3.  RE: PAW - Automatic feeders

    Posted 03/10/23 04:18 PM

    Hi Asgeir,

    We are not planning on automatically generating feeders in the Workspace modeling workbench.  Rather, our strategy is to eventually remove the requirement for feeders while continuing to support consolidations of rule derived cells.  Removing the requirement for feeders will allow us to expose modeling capabilities to users that would not consider themselves traditional model developers.  Instead of users writing rules, we could generate feederless rules based on visual mappings between cubes (think links from Performance Modeler).  Removing feeders would also have a significant positive impact on the performance characteristics of the database.  Consider how long it can take rules and feeders to compile on large complex cubes.  The feeder generation solution in Performance Modeler had limitations in terms of scalability.

    We are on board with your requirement to make Planning Analytics modeling easier!



    ------------------------------
    Stuart King
    IBM Planning Analytics Offering Manager
    ------------------------------



  • 4.  RE: PAW - Automatic feeders

    Posted 03/14/23 09:11 PM

    Thank you @STUART KING  This is really great news.  

    The drag/drop linking, and rule features are convenient in PM but for a little more complex rules, the handwritten code is better to understand. Writing rules with a function calling a function or attrs is easier in PAW than it was in PM. Also PM's ability to automatically generate summarized cubes and }attributes from links are convenient but can lead to maintenance problems later. Most people working with data are excellent in coding Excel functions and I guess those people would love to hand-code some rules and links in PAW and not to worry about the feeders. 

    Thanks again, Asgeir 



    ------------------------------
    Asgeir Thorgeirsson
    ------------------------------



  • 5.  RE: PAW - Automatic feeders

    Posted 07/29/26 02:02 PM

    Hi Stuart, Can you please confirm if this is in the Roadmap. Or something that is being considered in SaS.. Wanted to understand the timeline



    ------------------------------
    Amit Chakravorty
    ------------------------------



  • 6.  RE: PAW - Automatic feeders

    Posted 2 days ago

    Hi Asgeir,

    I came across this discussion and noticed Amit raised the roadmap / SaaS question again in July 2026, but I don't see a follow-up answer yet.

    What I find particularly interesting in Stuart's original response is that the proposed direction was not simply automatic feeder generation, but eventually removing the requirement for feeders while continuing to support consolidations of rule-derived cells.

    That distinction still seems important from a model developer's perspective. Today, we still work with the traditional Check Feeders / Process Feeders workflow, and there are scenarios in v12 where rule-derived cells require feeders for functionality such as MDX FILTER.

    So I would also be interested in a current update on the technical direction, rather than only the timeline:

    Is the feederless approach described in 2023 still being pursued for handwritten rules using SKIPCHECK, including more complex cases such as inter-cube DB() rules and consolidations of rule-derived cells?

    Or has the scope changed since the original statement, with feederless modeling primarily being considered for generated / visual mappings?

    For larger existing models, the distinction would be important. Better feeder generation would improve the current approach, while genuinely removing the feeder requirement would fundamentally change how we design and maintain rule-based models.

    If IBM has an updated view on this, I'd be very interested to hear it.



    ------------------------------
    Vitalij Rusakovskij
    ------------------------------