Abstract
Eventually, you will (hopefully) have developed and delivered something in TM1 that is unique enough (and adds significant value) that you’ll want to leverage it in other models.
Leverage Considerations
Before leveraging your component, the following must be considered:
- General Business Requirements (do your business requirements match the delivered components business specifications?)
- Component Mechanics (can the model provide the component its minimal input parameters?)
- Architectural Position (does the model adhere to the architectural principals required to support the component in its current state?)
General Business Requirements
Most importantly, you’ll need to determine if the “destination models” business specifications are defined as to adhere to or take advance of the specifications of the component you want to leverage. Does the component actually do what you need? To answer this question, you need to examine the inter-workings or “basic mechanics” of the component. The First step? Examine the components documentation. No documentation? Then you shouldn’t be considering leveraging it (it’s not mature enough).
Architectural Position
To effectively leverage your component, the destination model should adhere to basic industry architectural best practices. Although the basic logic of the component can be replicated as new TM1 objects (i.e. copy and paste), to produce a scalable, sustainable application, the following principals should be evaluated:
- Architectural Purity
- Data Absorption and Staging
- Configuration of Assumptions and Adjustments
- Results Consumption
Architectural Purity
All Models are made up of 4 distinct components: Absorption, Configuration, Calculation and Consumption. These components should be kept separate from your component logic. Although the component can be implemented as part of any of these existing components, for sustainability and scalability, it is not recommended.
If the component is to be implemented in an existing model, specific cubes should be utilized to keep the components logic encapsulated and decoupled from the existing model logic. For new models, the components logic can be “designed in” to the overall architecture.
Data Absorption and Staging
In an architecturally pure state, all data “absorption” of a model would be kept separate from the actual models “business purpose”. This would be accomplished through the use of “data staging” cubes and TM1’s ETL tool (TurboIntegrator) and/or the TM1 Rules Components. If this is not the case (typically when implementing the components into a previously deployed model), it is recommended that the data that will be absorbed by the leveraged component be staged and transformed before being absorbed by the leveraged component.
Configuration of Assumptions and Adjustments
Typically all models will require certain assumptions to be configured (placed in context). Models also allow users to make adjustments (to support various scenarios). The leveraged component will require certain assumptions and adjustments to be made and this processing must be kept exterior to the actual components logic.
It is recommended that the assumption and adjustment method used by the destination model be leveraged to capture the input of assumption parameters and data adjustments required by the leveraged component.
Hopefully, the leveraged component will use a series of simple lookup cubes (LK) to store assumption configurations and track user adjustments, and it is recommended that these cubes be leveraged in the destination model.
Results Consumption
The summarizing and compiling of the results of complex model calculations is considered a generic TM1 operation, not a key value-add to any particular model. Therefore the method and supporting logic to consume this information should be kept separate from the models “value-add” logic. Consuming the results of the leveraged component should also be kept separate from other components by utilizing supporting reporting or consumption cubes) or appropriate method).
Leverage On!