Dynamic Planning: Best Practices for Model Building, Model Refresh, and Model Restructure
  • 3 Minutes to read
  • Dark
    Light
  • PDF

Dynamic Planning: Best Practices for Model Building, Model Refresh, and Model Restructure

  • Dark
    Light
  • PDF

Article summary

Overview

This topic is intended for administrators and provides in-depth information on setting Key and Value dimension types in your model and incrementally refreshing the model to reduce the time required.

Model Building Best Practices

Identify the dimensions needed for analysis or reporting requirements. Then set them as Key or Value Type dimensions.

Model Building: Common Guidelines for Dimensions

Key type dimensions:

  • Growing dimensions, such as Products, Projects, Customer, Companies

  • Dimensions that need to be scoped, such as Scenario, Time, Measures

  • Dimensions that might be used in a formula reference

Value type dimensions:

  • Static dimensions in terms of size; they are not growing rapidly

  • Dimensions with a large number of levels (if possible, it could be defined as Attribute)

  • Departments, Regions dimensions

Model Building: Size Guidelines

Big Model

20GB

Records (Application Statistics)

8M to 10M blocks

Value Block Size

150K to 180K cells

 

Medium Model

12GB to 18GB

Records (Application Statistics)

4M to 8M blocks

Value Block Size

100K to 150K cells

 

Small Model

5GB to 10GB

Records (Application Statistics)

1M to 2M blocks

Value Block Size

60K to 100K cells

Model Building: Tips

  • If dimension names or types are incorrect after saving, you need to use "Modify Model" to make changes. You can either delete or modify the dimension type using this option.

  • You must have at least one Key and one Value dimension defined

  • The order of your dimensions in Model Setup will dictate how your dimensions are displayed in the Analyze task.

Model Refresh Best Practices

Perform a full refresh when building models and then switch to incremental refresh mode.

Full Refresh

  • Adjust rollup operators to ‘~’ in dimensions like Scenario, Time, Measures, Reporting as needed (one-time setup).

  • Run refresh by each Scenario and Year combination rather than running all scenarios in one step.

  • If you have QTD or YTD measures in your model, make sure you run a formula to get the expected results at the quarter level and year level after the Aggregation step.

Incremental Refresh

  • Identify the data set that has changed, such as Actual for the current period and prior period.

  • Only the changed data will be cleared and refreshed.

  • Add any new chart-of-account members to your model as part of your steps, such as adding new accounts. System defined dimensions like Scenario and Time would require different steps.

  • If you have QTD or YTD measures in your model, make sure you run a formula to get the expected results at the quarter level and year level after the Aggregation step.

Incremental Refresh – System-Defined Dimensions Process

In Practice: Adding a New Scenario to the Model:

  1. Add the Scenario under Model, Dimension.

  2. Run Metadata Download under Manage, Application Administration if you are missing the new scenario in the HACPM_Financial model under Model, Dimension.

  3. Update the Data Map to include the new scenario and then run the incremental refresh process for the new scenario.

In Practice: Adding a New Fiscal Year to the Model:

  1. Add the time member(s) for the new fiscal year under Model, Dimension. Use Append mode by copying the hierarchy from the HACPM_Financial model.

  2. Update the Calculation to use the new periods and then run the incremental refresh process.

Incremental Refresh: Tips

  • Set up the model refresh through the Scheduler to run overnight automatically.

  • Run test model refreshes and adjust the model for optimal performance before moving the model to Production for users.

Model Restructure

Whenever you modify a hierarchy, including adding or moving members or sub-hierarchies, Dynamic Planning performs a restructure operation. Depending on the model size and configuration, the operation can be a lengthy process (4 or more hours). We have improved the restructuring operation to be significantly faster when modifying the hierarchy of large dimensions and models. It is optimized to be completed in less than 4 hours. There is no additional configuration needed.

If the model is generated and you want to update the hierarchy, then simply lock the model and update the hierarchy.

You can optionally select to enable a Data Storage Partitioning enhancement. It is an opt-in feature and will be enabled only for selective customers based on the use case. This data storage partitioning enhancement splits the internal model storage and stores leaf-level and rollup-level data separately, thus improving the performance of some operations, such as running a hierarchy update or running a map. Contact Planful Support if you want to opt-in to this feature.


Was this article helpful?