SNP Heuristics background job is running for a long time

[former_member215376]

Dear Experts, Gurus,

Recently we had some additional locations and products in our system because of some new projects coming in. Their numbers (in total) is really not so significant but I would say not less also.

We are running our daily SNP heuristics background jobs and so far we had no issue on the run time. However, when the said additional locations and products came in, the issue now is that the background jobs seem not processing at all. Looking at the job log, it is always stuck at "0 Time Series generated" which is normal as we could also find this message in the previous successful runs.

We tried to run for small and big selection and we found no issue. This is excluding the additional locations and products.

We tried to run for the additional locations and products selection and this time we seemed to root that the issue was coming from them.

Questions:

     1 - We are using parallel processing and we believe that we have enough space to accommodate the additional locations and products, is there any other way that we could check how to identify the 'real' root cause? Our option is through debugging.

     2 - Did you have the same experience when some new projects coming in and you did some improvements in the SNP heuristics? What can be improved aside from parallel processing? Is it a good idea to have a separate steps/variants for SNP heuristics?

     3 - Is there any maximum limit of master data that can be selected in a single run?

Many thanks for the help and assist.

Best regards,

AA

[SatishWaghmare]

Hello AA

Here are a few things which comes to my mind -

1. It is recommended to run Low Level Code (/SAPAPO/SNPLLC) before running SNP Heuristic. This will also helps in identifying recursive T-lanes besides its benefit performance during the SNP run.

2. Run /SAPAPO/TS_LCM_CONS_CHECK on daily basis before the planning run.

3. It will also help if /SAPAPO/TS_LCM_REORG_SNP this is run periodically (may be once a week)

4. When SNP Heuristic, it goes through multiple steps - Reading Data, Solution Calculation, Order Creation and Transfer of the orders etc.  Any thing you can run in parallel will certainly help - using of parallel processing profile, and also by splitting the variant to run those in parallel etc.

5. Also refer below link if it is of any help.  I personally like Performance Monitor as it gives some good insight - which step within Heursitc is taking longer, Size of the memory used, No. of Demands, No. of Periods, No. Location Products, No of Planned Order / Pur Reqs created and so on.

Statistics - SNP Heuristic - APO Performance Monitor - SAP Library

Try accessing using APO Performance Monitor by choosing Advanced Planning and Optimization APO Administration APO Performance Monitor .

This is a great tool to analyse performance in other areas as well like BOP, Optimizer, CTM, DP-Runs etc.

Hope this will help.

Thank you

Satish Waghmare

[SatishWaghmare]

Hello AA

If your question is answered, SCN practice recommeds to close the thread by identifying the answer as correct/helpful.

Thank you

Satish Waghmare

[alexander_jankovsky]

Try to run a location heuristic in your SNP planning book - first for a single item - you will get an error message if there is a specific problem

[nitin_thatte]

There could be an issue with low level code generation.  There could be cycles in model.

Please check.

Nitin Thatte

[former_member187488]

Hi

Just a guess -- this may due to SNP master data update. Please refer to note 1331576.
You can execute report  /SAPAPO/TS_LCM_QUEUE_UPDATE and then report /sapapo/ts_lcm_cons_check before heuristic run and then see if the situation could be improved.

BR,
Ada