Planning Sequence for Bottom-Up Planning
Defining the Planning Sequence
We will investigate an example of a planning sequence that prepares a bottom-up planning process. It consists of 3 steps (see the figure).
...
1. A linear regression is computed for all existing products and regions based on the actual data for a number of sales items; this results in the data Forecast for the months of the following year. This version of the data is stored under the version PLAN_LIN.
2. The sales figures are derived for two new products from similar products. This is done with a copy function.
3. In bottom-up planning, the planners responsible for the individual regions should adjust the forecasted numbers to their estimates. However, since you want to get the forecast data for comparison purposes, an additional version of the data is provided for this planning session. This is also done with a copy function.
All three functions in this example work on the same aggregation level.
Characteristics: Version, Calendar Month, Region and Product.
Key figure: Number of items sold
All characteristics in all the realtime-enabled InfoProviders are relevant for locking and therefore can be used as packaging characteristic.
Overview of the Functions Contained in the Planning Sequence
Step |
1. |
2. |
3. |
Function Type |
Forecast |
Copy |
Copy |
Comments |
Writes the data for forecast period to version PLAN_LIN. Reference data is read from version ACTUALS. |
Copies data in version PLAN_LIN to two new products based on similar products that were already sold. |
Copies the forecast data from version PLAN_LIN to an additional working version PLAN_BOTTOM for the BOTTOM-UP planning. |
Filter |
Version = PLAN_LIN, Month = Jan 2007–Dec 2007 |
Version = PLAN_LIN |
Version = PLAN_BOTTOM |
Characteristics of Function to be Changed |
Month |
Product |
Version |
Parameters of Function |
Past: Jan 2005 – Dec 2006, Past filter: Version = ACTUALS, Forecast period: Jan 2007 – Dec 2007, Forecast parameter: Linear Regression |
Activity 1: From product 77 To product 105 Activity 2: From product 93 To product 106 |
Activity 1: From version PLAN_LIN To version PLAN_BOTTOM |
Scheduling the Planning Sequence
The planning sequence should be scheduled within a process chain.
The planning sequence is selected in the detailed maintenance of the process, but no variable variant is selected because all the contained variables can be replaced without user interaction.
Since the data volume is of medium size, the setting Process in packages is selected. For reasons of clarity, the steps in the example are divided into few packages, but very different packaging types are selected for these packages.
Dividing the Planning Sequence into Packages and Packaging Types
Step |
1. |
2. |
3. |
Type of packaging |
Configured |
No packaging |
Automatically |
InfoObject |
Region |
|
|
Method for dividing into packages |
Dimension |
|
Dimension |
Estimated number of values |
10 |
|
|
Package size |
4 |
|
|
Number of packages |
3 |
1 |
3 |
Processing
The following figure gives an overview of a master process that links the planning sequence into the process chain. It manages the execution of the planning sequence, schedules the subjobs, and collects the results of the subjobs. The filters of the steps are divided into packages according to the settings and executed in parallel background processes using the BI background management.
Master process step 1:
● The master collects all the filters and sets a write lock for the BI lock server. The BI lock server grants it a master lock. The real lock is then released.
● The first filter is divided into three packages with a maximum of four values with the predefined characteristic Region.
● Three subjobs are scheduled with the BI background management.
Master process step 2:
● The master waits until the BI background management has executed all three subjobs. The system only performs the next step when all the subjobs have been executed.
● No packaging is configured for step 2 of the planning sequence. For this reason, only one subjob is scheduled with the entire filter.
Master process step 3:
● The master waits until the BI background management has executed the subjob. The system only performs the next step when this subjob has been executed.
● Automatic packaging with three packages is defined for step 3. The master looks for a characteristic that has at least three characteristic values and accordingly breaks the filter down into three packages.
● Three subjobs are scheduled with the BI background management.
End of Master Process:
● The overall status of the planning sequence is recorded and the logs are written for the last time.
● The system removes the master lock.