TM1 Forum

Hello,
with Palo 2.5 what are your experiences/recommendations with push rules and (TM1 like) enterprise rules? When should push rules still be used? What are limiting factors for enterprise rules (data volume etc.)?

My application in scope will have a small data volume but intensive calculations on different levels of organizational entities.

Thanks in advance.

Marcus

Marcus Scherer wrote: with Palo 2.5 what are your experiences/recommendations with push rules and (TM1 like) enterprise rules? When should push rules still be used? What are limiting factors for enterprise rules (data volume etc.)?

My application in scope will have a small data volume but intensive calculations on different levels of organizational entities.

Hi Marcus,

"Push rules" (which is IMHO just marketing terminology from the time that PALO didn't provide server-side "enterprise" rules) are in fact calulation and loading routines, which will only work on base level (N:) cells (you could calculate on a consolidated level, but the result will be splashed to base level).
So for calculation on C: level you need enterprise rules. Also for dynamic modeling ("real-time what-if") you need the enterprise rules - although you could use Supervison Server (commercial component) to set cell-change triggers on cube areas and have the calculation done at cell change.
In my experience the entreprise rules work well on small data volumes, even when there is a lot of rules/calculations. Make sure you use markers (the PALO equivalent of TM1 Skipcheck&Feeders) as these can influence calculation time significantly (I've seen calculation last for over 20 minutes without markers, but within a second with markers).


Michel

As Michel says, Push rules are what products that don't have "proper" rules use as a workaround.

Having said that I do have a forecasting app that I built in TM1 (and which I am porting to Palo) that only ever uses push rules.

It only forecasts at the N level (another of Michel's salient points) and it has some statistical calculation which is simple to write in an Excel formaulae but which I wouldn't want to attempt in TM1 or Palo. Possible - just no fun to be had there.

The real reason why this works in this particular instance is that the UI is a complex set of data with varying dimensionality that needs a spreadsheet with multiple cube references in contiguous areas to lay it out effectively for the user. Hence we are de facto in Excel land anyway so might as well make use of the functionality.

The combination of Excel formulae and palo.setdata works well in this instance, but it is a VERY specific example. In most cases I would regard push rules as a poor substitute for server based rules.

In my experience of Palo, which is mostly limited to the intensive development of a relatively small planning model (200 product groups X 156 weeks x 4 versions x 30 measures x 3 cost types x 8 data types) rules work very effectively and the editor is pretty smart once you get used to it. I have successfully ported this mode from TM1 where the rules covered over 1,000 lines in the .xru sheet!

No elisanc yet though (but the presence of Undo and Undo Spread makes me forgive them a lot )