Calculation Performance

[KMG]

Hello guys, first question here on the forum...I'll try to be as clearer as possible in order to let no place for missunderstandings...

It is related to Tm1 performance.

We have a certain model, in which are involved: a single "central" cube, which holds all data and dimensions such as versions, hierarchy, measures etc, and several "Websheets" that are excel sheets accessed through TM1 Web. A couple of these websheets will be "Reports" generated by calculations done within the model.

The model will be used by around 30 people, and questions are:

1) Is it better in terms of performance to do calculations on cubes through rules or to just do them directly in the Excel sheets? For example, we have to decide if we are gonna make a calculation using an excel formula in the websheet or using rules in a cube for a particular element in the measures dimension and then connecting the websheet to that particular cell.

2) is it better in terms of performance to have only 1 excel file that generates the several reports changing through conditional formatting or just to make one for each report?

Thanks a lot and hope you can help us!

Regards,

Bruno

[jim wood]

1) Depends on the nature of the calculation. There are certain things like scaling your results in to thousands that TM1 can do but it would be more sensible to do in Excel.

2) I'm not 100% certain this will have an impact either way. I guess it depends on your network. TM1 is very "chatty" and the network traffic generated by a larger data set may take more time. Again this depends on the number of formulas in the excel workbook and the complexity of this one cube.

I know this may sound very general but without knowing your model in more detail it's hard to be very specific,

Jim.

[KMG]

1) Depends on the nature of the calculation. There are certain things like scaling your results in to thousands that TM1 can do but it would be more sensible to do in Excel.

2) I'm not 100% certain this will have an impact either way. I guess it depends on your network. TM1 is very "chatty" and the network traffic generated by a larger data set may take more time. Again this depends on the number of formulas in the excel workbook and the complexity of this one cube.

I know this may sound very general but without knowing your model in more detail it's hard to be very specific,

Jim.

Let me know what kind of details would you like to know in order to have a more specific response!!

Just to let you know, calculations are mostly like percentages, susbtractions, that kind of stuff.

Thanks a lot for your answer it actually helps a lot!

Thanks

Bruno

[jim wood]

I would do subtractions with the measure hierarchy. As for percentages I would do them within TM1. People have a nasty habit of getting them wrong in excel.

With the above and the point about scaling I made earlier you should have enough to get you started,

Jim.

[declanr]

People have a nasty habit of getting them wrong in excel.

Not to mention my (possibly over zealous) love of the fact that in tm1 when dividing by 0 you can use the " \ " symbol to return 0 as opposed to an error... I'm always amused by the simple things.

Like Jim has stated it all depends on YOUR model but generally speaking; well written rules with appropriate feeders would always be my preference over excel calculations even if only for the fact that you've already paid an arm and a leg for the product so you may as well use it fully!

[KMG]

I would do subtractions with the measure hierarchy. As for percentages I would do them within TM1. People have a nasty habit of getting them wrong in excel.

With the above and the point about scaling I made earlier you should have enough to get you started,

Jim.

Thanks a lot for your quick reply!

We'll start discussing here these things you mentioned.

Thanks again,

Bruno