Module with multiple dimentions
My company is now hiring an external consultant for anaplan model building. One thing I wanna understand more from the community is that when do we build different modules for calculations. So here is the original data
Let's say it's a table with Dimension ABCDE , and metrics FGHi
ABCDE could be list like Industry, Location, Company name
And the output is supposed to be pages where we have the aggregation of the metrics by different dimensions (or combination of dimentions)
Instead of building a module having all ABCDE dimentions together, the model builder is doing it in a duplicated way.
AB
AC
AD
AE
Her reason is that the dimeition of ABCDE in a module would **** up the model (10*10*10*10*10 cells at least and other cells in the base calculations)
However, this is going to make the maintenance of the dashboard very complicated. What's the best practice for these multi dimension summaries?
Also - when is a time we should put the dimension (ABCDE in this case) as a dimension vs as a line item (by referring to the list)
Thanks
Best Answers
-
@GAPER To find if you have Polaris open a module in Blue Print. If you see the columns below highlighted in yellow then you have Polaris. Have a look the beauty of Polaris decreasing space use below. Each line item was 12 billion cells. I had 3 line items that I was really using totalling 36 billion cells but the model was under 1Gb. If you had classic this one module would have needed 300Gb easy. If you do have Polaris and your consultant is progressing with those composite lists you may need to look at a new consultant (like my company lol). Polaris is relatively new and am of the impression very few companies has started using it.
Also, with Polaris the number of cells will still increase with every dimension you add to it but it's really about how much memory it uses which is shown in the screenshot below.
0
Answers
-
@GAPER In classic model, I would agree with your consultant. In general you don't want to have several large dimensions intersecting in a single module. E.g. customers, products, location as separate dimensions in a single module as it would chew up space. So you create composite hierarchies. For example,
you would have a list hierarchy such as
- Region
- Country
- Area
- Customer
- Product
- Customer
- Area
- Country
But if you were using Polaris, you can go nuts with the amount of dimension intersections. But Polaris is best suited for data with high levels of sparsity. Plus the license is pricey. I have found models that spans global geographic regions easily satisfies this criteria
I suggest you watch this 10 minute Anaplan lesson on hierarchy design
0 - Region
-
I think we are using polaris (by the way how do we check if it's a classic model or a polaris?)
But the number of cells are increasing by every dimention i am adding to it
(like A*B*C*D*E) which quickly growing into a model of multiple gigabytes.
0 -
Thanks Tristan - I think it's a classical model then. How do we enable Polaris? Or do i get to choose that when i create a new model?
0 -
I see… Didn't know this could be another tier of charges… (Feel like the classic is a inferior product in this sense…)
Do you have an idea on how much more expensive and whether there is a way for an old model to transfer into a polaris enabled model?
Thanks
0 -
@FrankPau Polaris is like an add on you purchase which has $500K RRP on top of the usual licensing arrangement. So when purchasing an Anaplan license you first either either get a "Basic", "Professional" or "Enterprise" edition license. Like all commercial arrangements I believe Polaris licensing costs can be negotiated so best to discuss with your Anaplan contact.
1