TM1 Contributor

[dharav]

> TM1 10.2 Environment
> We have 7 views inside it.
View 1 from ABC cube, View 2&3 from PQR cube, View 4 from CDE cube and View 5,6,7 from XYZ Cube

> Problem: The performance of the View 1 is very slow. It takes more than 7 seconds to change data from one version to another version. It takes almost 2 minutes to take ownership.

Information on cube:

This cube is a source cube of Cube PQR.

The version dimension is a part of this cube and it also a control dimension. Manual Forecast, the element of the version dimension only has write rights and rest of them have Read rights in control dimension.

The rules in this cube are less than cube xyz. Most of the rules between these two cubes are similar. (The performance of the XYZ cube is much faster - taking hardly a second to move from one version to another version). Highest data and Calculations are in the cube XYZ Cube.

Could you please let me know the best solution of it. Let me know if you require any further information on this.

[jim wood]

Did you create the rules using performance modellor? The reason I ask is that the feeders generated by PM over feed. You may need to go in to architect and check the feeders. If this isn't the case you may need to provide more info,

Jim.

[dharav]

Hi Jim

I have created links using performance modler between Cube ABC and Cube PQR. Where, Cube ABC is a source cube and Cube PQR is a target cube.
ABC CUBE
Dim1: Regions
Dim2: Forecast Version
Dim3: Accounts
Dim4: Month
PQR Cube
Dim1: Regions
Dim2: Versions (same element as forecast version dim of cube ABC)
Dim3: Accounts
Dim4: Months Total (Same elements as month dim of cube ABC)


################Automated Generated Rule in the cube PQR (Link based rule)
#Region Link rule: ABC --> PQR - String
#Source cube: ABC
#Target cube: PQR
#Autogenerated LINK STRING 7D4C696E6B5F4163636F756E747320496E707574202D2D5F334520466F7265636173742050264C
#@AutoGenerated
['Version':{'Nov (10+2) Forecast','Actual - Current Year','Next Yr Prelim Plan','YearEnd (12+0)','Previous Fcast','May (4+8) Forecast','Apr (3+9) Forecast','Sep (8+4) Forecast','Jul (6+6) Forecast','Current Yr Plan Detail','Actual - Prior Prior Year','Mar (2+10) Forecast','Oct (9+3) Forecast','Aug (7+5) Forecast','Dec (11+1) Forecast','Jun (5+7) Forecast','Actual - Prior Year','Feb (1+11) Forecast','Jan (0+12) Forecast'}] = S:IF(DIMIX('Month',
!1t_MonthsTotal) <> 0, DB('ABC', !regions,
ATTRS('Version', !Version, '}Map_}Link_ABC --_3E PQR'),
!Accounts, !Months Total), CONTINUE);
#EndRegion

#Region Link rule: ABC --> PQR - Numeric
#Source cube: ABC
#Target cube: PQR
#Autogenerated LINK NUMERIC 7D4C696E6B5F4163636F756E747320496E707574202D2D5F334520466F7265636173742050264C
#@AutoGenerated
['Version':{'Nov (10+2) Forecast','Actual - Current Year','Next Yr Prelim Plan','YearEnd (12+0)','Previous Fcast','May (4+8) Forecast','Apr (3+9) Forecast','Sep (8+4) Forecast','Jul (6+6) Forecast','Current Yr Plan Detail','Actual - Prior Prior Year','Mar (2+10) Forecast','Oct (9+3) Forecast','Aug (7+5) Forecast','Dec (11+1) Forecast','Jun (5+7) Forecast','Actual - Prior Year','Feb (1+11) Forecast','Jan (0+12) Forecast'}] = N:IF(DIMIX('Month',
!Months Total) <> 0, DB('Accounts', !regions,
ATTRS('Version', !Version, '}Map_}Link_ABC --_3E PQR'),
!Accounts, !Months Total), CONTINUE);
#EndRegion


############## Auto Generated Feeders in the Cube ABC ###############

['Forecast Version':'Oct (9+3) Forecast'] => DB('1r_ForecastP&L', !Regions, 'Oct (9+3) Forecast',
!Accounts, !Month);
['Forecast Version':'Actual - Prior Year'] => DB('1r_ForecastP&L', !Regions, 'Actual - Prior Year',
!Accounts, !Month);
['Forecast Version':'Jun (5+7) Forecast'] => DB('1r_ForecastP&L', !Regions, 'Jun (5+7) Forecast',
!Accounts, !Month);
['Forecast Version':'Dec (11+1) Forecast'] => DB('1r_ForecastP&L', !Regions, 'Dec (11+1) Forecast',
!Accounts, !Month);
['Forecast Version':'May (4+8) Forecast'] => DB('1r_ForecastP&L', !Regions, 'May (4+8) Forecast',
!Accounts, !Month);
['Forecast Version':'Current Year Plan V2'] => DB('1r_ForecastP&L', !Regions, 'Current Yr Plan Detail',
!Accounts, !Month);
['Forecast Version':'Aug (7+5) Forecast'] => DB('1r_ForecastP&L', !Regions, 'Aug (7+5) Forecast',
!Accounts, !Month);
['Forecast Version':'Next Yr Prelim Plan'] => DB('1r_ForecastP&L', !Regions, 'Next Yr Prelim Plan',
!Accounts, !Month);
['Forecast Version':'Jan (0+12) Forecast'] => DB('1r_ForecastP&L', !Regions, 'Jan (0+12) Forecast',
!Accounts, !Month);
['Forecast Version':'Sep (8+4) Forecast'] => DB('1r_ForecastP&L', !Regions, 'Sep (8+4) Forecast',
!Accounts, !Month);
['Forecast Version':'Mar (2+10) Forecast'] => DB('1r_ForecastP&L', !Regions, 'Mar (2+10) Forecast',
!Accounts, !Month);
['Forecast Version':'Apr (3+9) Forecast'] => DB('1r_ForecastP&L', !Regions, 'Apr (3+9) Forecast',
!Accounts, !Month);
['Forecast Version':'Actual - Prior Prior Year'] => DB('1r_ForecastP&L', !Regions, 'Actual - Prior Prior Year',
!Accounts, !Month);
['Forecast Version':'Nov (10+2) Forecast'] => DB('1r_ForecastP&L', !Regions, 'Nov (10+2) Forecast',
!Accounts, !Month);
['Forecast Version':'Jul (6+6) Forecast'] => DB('1r_ForecastP&L', !Regions, 'Jul (6+6) Forecast',
!Accounts, !Month);
['Forecast Version':'Feb (1+11) Forecast'] => DB('1r_ForecastP&L', !Regions, 'Feb (1+11) Forecast',
!Accounts, !Month);
['Forecast Version':'Actual - Current Year'] => DB('1r_ForecastP&L', !Regions, 'Actual - Current Year',
!Accounts, !Month);
['Forecast Version':'Previous Fcast'] => DB('1r_ForecastP&L', !Regions, 'Previous Fcast', !Accounts,
!Month);
['Forecast Version':'YearEnd (12+0)'] => DB('1r_ForecastP&L', !Regions, 'YearEnd (12+0)', !Accounts,
!Month);


We have approx 3400 elements in the region dim and 400 elements in the account dim
Is there any opportunities to improve feeders further more or write a common rule instead of Links?
Could i use following rule instead of links? I know that overfeeding will be there but could you let me know the best alternative?
Rule Inside the PQR Cube
[]=db('ABC',!regions,!Version,!Accounts,!Month Total);
Feeder inside the ABC Cube
[]=>db('PQR',!regions,!Forecast Version,!Accounts,!Month);

[jim wood]

While PM is a decent tool for those new to TM1 (especially those who know EP) but it does have it's limitations. One of them is feerders. It always over feeds to avoid values being missed, which is pretty understandable given the circumstances. I don't however feel that this is your problem. Is there a reason why you've seperated this in to so many cubes? Keep in mind that if you did used to use EP, TM1 doesn't suffer from it's size limitations so there's no need to split your model in to so many cubes. There may however be a reasonable reason why you've split them. Also do the links need to be rule based? Can some of the links be TI based to save on the memory / performance overhead?

[dharav]

I have set up these links as data should be updated as input data enter by the users. Due to which i can not use TI here.

Could you let me know, which one is better? The Link that i have created or the rules that i have mentioned. I would definitely try the rules one today and check the performance.

If you have any suggestion on either rules or the links which could enhance the performance than please recommend those

Thanks

Dharav

[declanr]

To take this from a different perspective; what is the position of your approval dimension in each of the 2 cubes (i.e. near the beginning or near the end)?

I haven't personally noticed any performance hit yet but when saving an application it throws up a warning message saying that performance may suffer if the position of your approval dimension in a given cube is near the bottom.

[jim wood]

Good point Declan, cube optimization should help. If that doesn't help we can always revisit rules and feeders in more detail,

Jim.

[dharav]

Hello

The approval hierarchy is the subset of region dim. Region dim has been used on top (first dim) in both cubes ABC and PQR respectively.