-
Type:
Feature Request
-
Status: Closed (View Workflow)
-
Priority:
Major
-
Resolution: Completed
-
Affects Version/s: None
-
Fix Version/s: CFA 2019R2
-
Component/s: DLPS, Forecasting
The following schedule of jobs is configured in Production:
- Midnight - DLPS generation
- 2 am - Forecast Import
- 4 am - InFORM Forecast generation
The original goal of the forecast import was to complete in 2 hours, the same time as the InFORM Forecast generation.
However, the forecast export is taking longer than we would like, typically completing within an hour, but only for 150+ stores.
Upon investigation, it has been determined that both the DLPS generation job and Forecast Import process place high demands on the CPU of the app server.
This feature is to investigate options to reduce this demand, and look at options to increase the speed at which forecasts are imported, while still completing the DLPS generation in the 00:00-04:00 window.
Example of Forecast Import processing:
- Today (2/14) the forecast import processed 800+ files in 2.5 hours
- Starting at 02:00
- Finishing at 04:32
- A typical day should ultimately process 4,000 files within a given time range (our original goal was 2 hours, but this may be a little bit ambitious)
- Note: there is work underway to process 50% of the files as summary files, which only contain data for the Metric Details, the Sales Item Detail and InventoryItemDetail sections will be blank
- is implemented by
-
CFAMX-2412 Documentation
- Closed
-
CFAMX-2419 SPIKE - Research DLPS process with a view to reducing the CPU utilization
- Closed
-
CFAMX-2424 SPIKE - Research the Forecast Import with a view to reducing the CPU utilization
- Closed
-
CFAMX-2675 Deploy and Test DLPS Optimization
- Closed
-
CFAMX-2717 Investigate a method to trickle the Forecast Import
- Closed
-
CFAMX-2933 Evenly spread processing Queue Messages over a period time
- Closed