How to set the optimal number of sessions for MRP parallel processing

[Former Member]

One of the features within SAP’s MRP functionality is the ability to assign processing across multiple servers or sessions to improve the runtime performance of the planning run. This feature is called parallel processing and is accessed through configuration using OMIQ to define parallel processing in MRP. 

To figure out how to determine the optimal number of sessions for parallel processing, we need to understand how this works.  When using parallel processing, it will execute in terms of packages that use the low level code.  This low level code is a component of the bill of material structure which we can use as a guideline for this.  A bill of material is broken down by low level codes for each of the BOM levels.  Determine what the deepest bill of material is and use this as your baseline number for low level codes.  If the deepest bill of material is 3 levels deep, then you have low level codes 000 (end item), 001,
002 for your BOM levels.  Thus your baseline number should be 3.

Since reorder point materials are not planned by requirements, these types of materials will be assigned a low level code 999 by the MRP program.  That adds one more to your number for sessions.  Now we have the 3 BOM levels and 1 reorder point level equals a total of 4 sessions.   In example below, I have configured 4 sessions for parallel processing.

Each low level code is a work package that once completed; the next package will begin processing.  Therefore if you have multiple sessions configured, then multiple work packages will be distributed over this number with a fixed number of materials assigned by the MRP program.  When you set your MRP planning variant or run directly using MD01, you have an indicator that you may now set for parallel processing.  This indicator will set the planning run to check the configuration setting for how many sessions to create.  In my example, it was set to 4.

After the MRP run, you can then see where the low level codes are identified for the materials and at the end of the statistics shows how many material planned for each session.  This method provides you a quick and easy way to determine the appropriate configuration level for using MRP Parallel processing.