Planning Analytics

hi,

i have a client who have some cube with a lot of rules with 1 big dimension (2 millions elements) and some other small dimension.

the problem today is i have to wait at least 5 hours for save rules if i change something in rule editor (and for make new rule or make a correction, i have to test a lot)

any tips to reduce the waiting time of save rule in this case? :?

thank a lof for help.

same post in TM1 forum : https://www.tm1forum.com/viewtopic.php?t=16320
------------------------------
Duong DAU
Consultant EPM Finance Transformation & Performance - Micropole


M +33 (0)7 62 61 85 33
F +33 (0) 4 26 99 60 81
131 Boulevard Stalingrad / 69100 Villeurbanne
------------------------------
#PlanningAnalyticswithWatson

Hi Duong,

Without knowing any more about the cube and the rules it's hard to give recommendations as we don't know if the calculation(s) need to be dynamic/ instant. Are the rules pulling data from multiple cubes? and whether other efficiencies could be found.

- Can the rules be switched to a TI process?
- Do the rules need to apply for all 2 million elements?
- Are you overfeeding? Do you need too?

In terms of testing, if I have a large rule file to save then I tend to adjust the rules on a Development instance for a selected element/ group of elements to prove the rule change works before then applying on a larger scale (ideally in the development environment) before then migrating to production.

------------------------------
Edward Stuart
------------------------------

Original Message:
Sent: Thu October 06, 2022 05:02 AM
From: Duong DAU
Subject: rule save issue

hi,

i have a client who have some cube with a lot of rules with 1 big dimension (2 millions elements) and some other small dimension.

the problem today is i have to wait at least 5 hours for save rules if i change something in rule editor (and for make new rule or make a correction, i have to test a lot)

any tips to reduce the waiting time of save rule in this case? :?

thank a lof for help.

------------------------------
Duong DAU
Consultant EPM Finance Transformation & Performance - Micropole


M +33 (0)7 62 61 85 33
F +33 (0) 4 26 99 60 81
131 Boulevard Stalingrad / 69100 Villeurbanne
------------------------------

#PlanningAnalyticswithWatson

Hi Stuart, 

Thank you for your reponse, 

- There are a lot of data from multiple cube and a lot of IF condition
- Can the rules be switched to a TI process? ==> My client dont want to click an button, they want something dynamique
- What you mean when you said "apply for"? what technique that you talk about? 
- This is a old client that have a lot of rule in a cube, then to be honest, i dont know how the feeding work in this cube. 

In terms of testing, if I have a large rule file to save then I tend to adjust the rules on a Development instance for a selected element/ group of elements to prove the rule change works before then applying on a larger scale (ideally in the development environment) before then migrating to production.
==> i can not do that because the calcul is from a lot of element, then each element is calculed by another element ==> this is a good idea but it is hard to do with this client. 

is remove feeder help to save rule faster? 


Thank a lot!

------------------------------
Duong DAU
Consultant EPM Finance Transformation & Performance - Micropole


M +33 (0)7 62 61 85 33
F +33 (0) 4 26 99 60 81
131 Boulevard Stalingrad / 69100 Villeurbanne
------------------------------

Original Message:
Sent: Thu October 06, 2022 05:32 AM
From: Edward Stuart
Subject: rule save issue

Hi Duong,

Without knowing any more about the cube and the rules it's hard to give recommendations as we don't know if the calculation(s) need to be dynamic/ instant. Are the rules pulling data from multiple cubes? and whether other efficiencies could be found.

- Can the rules be switched to a TI process?
- Do the rules need to apply for all 2 million elements?
- Are you overfeeding? Do you need too?

In terms of testing, if I have a large rule file to save then I tend to adjust the rules on a Development instance for a selected element/ group of elements to prove the rule change works before then applying on a larger scale (ideally in the development environment) before then migrating to production.

------------------------------
Edward Stuart

Original Message:
Sent: Thu October 06, 2022 05:02 AM
From: Duong DAU
Subject: rule save issue

hi,

i have a client who have some cube with a lot of rules with 1 big dimension (2 millions elements) and some other small dimension.

the problem today is i have to wait at least 5 hours for save rules if i change something in rule editor (and for make new rule or make a correction, i have to test a lot)

any tips to reduce the waiting time of save rule in this case? :?

thank a lof for help.

------------------------------
Duong DAU
Consultant EPM Finance Transformation & Performance - Micropole


M +33 (0)7 62 61 85 33
F +33 (0) 4 26 99 60 81
131 Boulevard Stalingrad / 69100 Villeurbanne
------------------------------

#PlanningAnalyticswithWatson

The assumption is that the rule is applying for all elements in the cube, in this instance you have a lot of elements (2 million+) 

Is it possible to restrict the rules to apply for only a selection of the cube? 

For example you may be calculating rules for all historical periods? If this is the case then you could restrict the rules to apply only for the Current Period (if this fits the rule requirements) and use a TI to populate historical periods.

Alternatively, you may be using a single catchall rule/ feeder across all 2 million elements, whereas only 500,000 need the rule/ feeder applying. This could mean write additional rules/ feeders with more specific targeting but is the balance between code maintainability and performance.

If you are unsure of how the feeders work I would make it a priority to investigate. Updating rules when unsure of the feeders is a risk that should be avoided.

Feeders are an essential component (until Feeder-less rules becomes a reality) of building IBM Planning analytics models. It is always worth refreshing your knowledge with the many resources available:

https://www.ibm.com/docs/en/planning-analytics/2.0.0?topic=rules-improving-performance-feeders
https://www.ibm.com/support/pages/tm1-server-feeders-guidelines-and-best-practices

------------------------------
Edward Stuart
------------------------------

Original Message:
Sent: Thu October 06, 2022 05:43 AM
From: Duong DAU
Subject: rule save issue

Hi Stuart, 

Thank you for your reponse, 

- There are a lot of data from multiple cube and a lot of IF condition
- Can the rules be switched to a TI process? ==> My client dont want to click an button, they want something dynamique
- What you mean when you said "apply for"? what technique that you talk about? 
- This is a old client that have a lot of rule in a cube, then to be honest, i dont know how the feeding work in this cube. 

In terms of testing, if I have a large rule file to save then I tend to adjust the rules on a Development instance for a selected element/ group of elements to prove the rule change works before then applying on a larger scale (ideally in the development environment) before then migrating to production.
==> i can not do that because the calcul is from a lot of element, then each element is calculed by another element ==> this is a good idea but it is hard to do with this client. 

is remove feeder help to save rule faster? 


Thank a lot!

------------------------------
Duong DAU
Consultant EPM Finance Transformation & Performance - Micropole


M +33 (0)7 62 61 85 33
F +33 (0) 4 26 99 60 81
131 Boulevard Stalingrad / 69100 Villeurbanne

Original Message:
Sent: Thu October 06, 2022 05:32 AM
From: Edward Stuart
Subject: rule save issue

Hi Duong,

Without knowing any more about the cube and the rules it's hard to give recommendations as we don't know if the calculation(s) need to be dynamic/ instant. Are the rules pulling data from multiple cubes? and whether other efficiencies could be found.

- Can the rules be switched to a TI process?
- Do the rules need to apply for all 2 million elements?
- Are you overfeeding? Do you need too?

In terms of testing, if I have a large rule file to save then I tend to adjust the rules on a Development instance for a selected element/ group of elements to prove the rule change works before then applying on a larger scale (ideally in the development environment) before then migrating to production.

------------------------------
Edward Stuart

Original Message:
Sent: Thu October 06, 2022 05:02 AM
From: Duong DAU
Subject: rule save issue

hi,

i have a client who have some cube with a lot of rules with 1 big dimension (2 millions elements) and some other small dimension.

the problem today is i have to wait at least 5 hours for save rules if i change something in rule editor (and for make new rule or make a correction, i have to test a lot)

any tips to reduce the waiting time of save rule in this case? :?

thank a lof for help.

------------------------------
Duong DAU
Consultant EPM Finance Transformation & Performance - Micropole


M +33 (0)7 62 61 85 33
F +33 (0) 4 26 99 60 81
131 Boulevard Stalingrad / 69100 Villeurbanne
------------------------------

#PlanningAnalyticswithWatson

Thanks Stuart, 

This is my first client who want that big cube. i never see this before. i will make a deep analyze to improve the feeder !

------------------------------
Duong DAU
Consultant EPM Finance Transformation & Performance - Micropole


M +33 (0)7 62 61 85 33
F +33 (0) 4 26 99 60 81
131 Boulevard Stalingrad / 69100 Villeurbanne
------------------------------

Original Message:
Sent: Thu October 06, 2022 06:27 AM
From: Edward Stuart
Subject: rule save issue

The assumption is that the rule is applying for all elements in the cube, in this instance you have a lot of elements (2 million+) 

Is it possible to restrict the rules to apply for only a selection of the cube? 

For example you may be calculating rules for all historical periods? If this is the case then you could restrict the rules to apply only for the Current Period (if this fits the rule requirements) and use a TI to populate historical periods.

Alternatively, you may be using a single catchall rule/ feeder across all 2 million elements, whereas only 500,000 need the rule/ feeder applying. This could mean write additional rules/ feeders with more specific targeting but is the balance between code maintainability and performance.

If you are unsure of how the feeders work I would make it a priority to investigate. Updating rules when unsure of the feeders is a risk that should be avoided.

Feeders are an essential component (until Feeder-less rules becomes a reality) of building IBM Planning analytics models. It is always worth refreshing your knowledge with the many resources available:

https://www.ibm.com/docs/en/planning-analytics/2.0.0?topic=rules-improving-performance-feeders
https://www.ibm.com/support/pages/tm1-server-feeders-guidelines-and-best-practices

------------------------------
Edward Stuart

Original Message:
Sent: Thu October 06, 2022 05:43 AM
From: Duong DAU
Subject: rule save issue

Hi Stuart, 

Thank you for your reponse, 

- There are a lot of data from multiple cube and a lot of IF condition
- Can the rules be switched to a TI process? ==> My client dont want to click an button, they want something dynamique
- What you mean when you said "apply for"? what technique that you talk about? 
- This is a old client that have a lot of rule in a cube, then to be honest, i dont know how the feeding work in this cube. 

In terms of testing, if I have a large rule file to save then I tend to adjust the rules on a Development instance for a selected element/ group of elements to prove the rule change works before then applying on a larger scale (ideally in the development environment) before then migrating to production.
==> i can not do that because the calcul is from a lot of element, then each element is calculed by another element ==> this is a good idea but it is hard to do with this client. 

is remove feeder help to save rule faster? 


Thank a lot!

------------------------------
Duong DAU
Consultant EPM Finance Transformation & Performance - Micropole


M +33 (0)7 62 61 85 33
F +33 (0) 4 26 99 60 81
131 Boulevard Stalingrad / 69100 Villeurbanne

Original Message:
Sent: Thu October 06, 2022 05:32 AM
From: Edward Stuart
Subject: rule save issue

Hi Duong,

Without knowing any more about the cube and the rules it's hard to give recommendations as we don't know if the calculation(s) need to be dynamic/ instant. Are the rules pulling data from multiple cubes? and whether other efficiencies could be found.

- Can the rules be switched to a TI process?
- Do the rules need to apply for all 2 million elements?
- Are you overfeeding? Do you need too?

In terms of testing, if I have a large rule file to save then I tend to adjust the rules on a Development instance for a selected element/ group of elements to prove the rule change works before then applying on a larger scale (ideally in the development environment) before then migrating to production.

------------------------------
Edward Stuart

Original Message:
Sent: Thu October 06, 2022 05:02 AM
From: Duong DAU
Subject: rule save issue

hi,

i have a client who have some cube with a lot of rules with 1 big dimension (2 millions elements) and some other small dimension.

the problem today is i have to wait at least 5 hours for save rules if i change something in rule editor (and for make new rule or make a correction, i have to test a lot)

any tips to reduce the waiting time of save rule in this case? :?

thank a lof for help.

------------------------------
Duong DAU
Consultant EPM Finance Transformation & Performance - Micropole


M +33 (0)7 62 61 85 33
F +33 (0) 4 26 99 60 81
131 Boulevard Stalingrad / 69100 Villeurbanne
------------------------------

#PlanningAnalyticswithWatson