Loading Data from Files
When history and initial seed data comes from flat files, use the following tasks to upload them into RAP:
Table 3-4 Flat File Load Overview
Activity | Description |
---|---|
Initialize dimensional data (products, locations, and so on) to provide a starting point for historical records to join with. Separate initial load processes are provided for this task. |
|
Run history loads in one or multiple cycles depending on the data volume, starting from the earliest date in history and loading forward to today. |
|
Reload dimensional data as needed throughout the process to maintain correct key values for all fact data. Dimensional files can be provided in the same package with history files and ad hoc processes run in sequence when loading. |
|
Seed initial values for positional facts using full snapshots of all active item/locations in the source system. This must be loaded for the date prior to the start of nightly batches to avoid gaps in ongoing data. |
|
Nightly batches must be started from the business date after the initial seeding was performed. |
Completing these steps will load all of your data into the Retail Insights data model, which is required for all implementations. From there, proceed with moving data downstream to other applications as needed, such as AI Foundation modules and Merchandise Financial Planning.
Note:
All steps are provided sequentially, but can be executed in parallel. For example, you may load dimensions into RI, then on to AI Foundation and Planning applications before loading any historical fact data. While historical fact data is loaded, other activities can occur in Planning such as the domain build and configuration updates.