MFGPRO Performance Problems

[nineinch]

facts:
Hyper-V
RH Enterprise Linux 5
OE10.2B
QAD2010SE

Start parameters:
-B 125000
-bibufs 100
-aibufs 150
-c 350
-L 80000
-Mf 60
-n 300
-spin 60000

6 instances of proapw
1 instance of probiw
1 instance of proaiw
2 instances of prowdog

200 active users

Problem:
During Net Change MRP performance problems for all users in same database, not in OS and not in other databases.
Up to 10 seconds frozen screens.
What can we do about these problems?

 

[mosfin]

try to read the QAD online help for menu 23.1 (Net Change MRP)
here is part of the help text:

Use Net Change Materials Plan to calculate demand and plan supply for items that have changed since material requirements planning (MRP) was last run.

MRP lets you plan supply to meet demand for purchased and manufactured items in a time-phased manner. It uses a site's master schedule and all other sources of demand and supply to:

o Calculate gross requirements and projected on-hand inventory

o Schedule and plan orders

o Produce action messages for managing the materials plan



This transaction may take some time to process. Submit it in batch to continue using your system for other tasks.

To shorten the processing time for generating a new plan using net change or regenerative MRP, consider changing the MRP horizon. This can be set temporarily in MRP Control. Make sure to reset it when MRP completes.

IMPROVED MRP PROCESSING

If you use Symmetric Multiple Processor (SMP) computers, you can use the Synchronized Calculation and Synchronization Code fields in MRP functions to run multiple MRP sessions concurrently. On SMP machines, you can run one session on each available machine for a given synchronization code against a single domain.

Additionally, if you have SMP computers and the Progress application server (AppServer), you can define AppServer instances that use multithreading to distribute the processing load among multiple processors. Specify AppServer setup information in MRP Control. You can update the default number of threads used each time you run MRP.