Question

CRP run time

  • 12 November 2020
  • 4 replies
  • 296 views
A
Rank
Userlevel 5
Badge +9

We have around 10 sites all running in same database.

All jobs like MS, MRP and CRP needs to be run daily due to our line of business.

We now run in to issues that the “night window” is getting  too small and jobs are not ready in time in the morning.

Especially the CRP is taking very long time to run for some sites, up till 90 minutes.

Does anybody know what is impacting the CRP execution time the most?

Is it the number of workcenters in use,  or the number of shoporder requisitions in the system (master schedules set to create sp for entire horizon) or a combination.

Any advise on this would be appreciated.

 

 

4 replies

M
Rank
Userlevel 6
Badge +12

Hello!

Do you run the job in sequence? Like for example:

MS for Site 1, MRP for site 1, CRP for site 1, MS for site 2, MRP for site 2, CRP for site 2 etc? Or do just run them as parallell chains of jobs? I mean, if  you run the planning jobs for site 1 parallell with the planning jobs for 2, etc.

Do you have the patch for the LCS case  G1975814-A ? 

In general the CRP performance is a function of:

  • Number of Shop Order Reqs + DOP Orders + Proposed Master Schedule Receipts
  • Number of operations per part
  • The length of the operations, well if they exceeds one day

I wish there were more functionality around our calendars so that you could remove past due calendar data - because the size of the calendar data tables are considerable. And the CRP calculation is using the calendar data soooo much.

-Mats

A
Rank
Userlevel 5
Badge +9

Hi Mats,

Thanks for your answer. 

We run the jobs in sequence, all planning jobs run in one queue with only one process. Do you mean setting the queue to run with 2 processes so we there can be 2 MS’s running in parallell?

We run on IFS 10 upd 3 and not sure G1975814-A is part of that, we have not had any CRP related patch installed, will ask support for that patch.

/Anna

M
Rank
Userlevel 6
Badge +12

Hej Anna!

MS, RRP, MRP/PMRP and CRP, they all use “collition job detection logic” at the beginning when they are about to startup. And what does that mean? Well, for example when you run MS Level 1 calculation on site X, system is checking:

  • Is there already an MS Level 1 or MS Level 0 calculation that is executing on site X?
  • Is there already an RRP calculation that is executing on site X?
  • Is there already an MRP/PMRP calculation that is executing on site X?
  • Is there already a CRP calculation that is executing on site X?

If ONE yes above, then the new MS Level 1 calculation on site X will not start, you will get an error in Background Jobs.

***

Back to your question - you sould be able to parallelize this because you are running the jobs on different sites.

BUT let say MS Level 1 calculation takes pretty long time for site X and the MS Level 1 caculation takes short time on site Y, then there is a risk that next job, let say MRP for site X starts to run, but it will be stopped because of MS Level 1 calculation on site X is still executing.

I have seen customer that have setup job chains - perhaps you should consider this. And I guess there are other possibilites as well.

I am at the “Rnd side”, consulting organisation and of course customers are very clever when it comes to this.

-Mats

J
Rank
Userlevel 4
Badge +9

Hey Anna,

We had the same issues. Already in IFS 8 we setup a job chain for all the planning funtions.

one job chain schedule per site  was assigned to a different job queue per site (with a small customization). This can be done in IFS 10 with standard functionality. 

Every queue was setup to execute only one process in parallel.

Hint: Make sure, that MRP is started after midnight. otherwise, MRP will create some strange results if started before midnight and ending after midnight.

/ Johannes

Reply


Terms & ConditionsCookie settings