Rescheduling runs slower in HANA

[m_olson]

We're in the middle of upgrading to Business Suite on HANA, and we were disappointed to see rescheduling (SDV03V02) actually take longer on HANA.

On a given night we reschedule approximately 50,000 line items, which is about 10,000 orders.

I found BADI SDV03V02_DBSYS_OPT, for which SAP provided a HANA-optimized implementation (SDV03V02_HDB_OPT).  Great I thought, this will be the answer.  I activated it, but still no noticeable improvement. The query to grab relevant items to be rescheduled runs very fast, but the actual processing is slower than our old system.

Function module SD_BACKORDER_CHECK_AND_SAVE is the culprit because I can see in debugging it reaches that FM very quickly, but then takes a long time to come back out of it.  I would think comparing apples-to-apples, running rescheduling on HANA (ECC 6.0, Release 7.4, Ehp 13) would be quicker than ECC 6.0 Release 7.02, Ehp 7.

Has anyone ran into any similar problems?

[dsadm]

Hi Matt,

we´re at the same point as you some weeks ago!

Did you already find a solution for yourself and want to share?

The activation of the BADI because of performance issues isn´t necessary as the ELSEIF sends the HDB into the same method.

IF lb_sdv03v02_dbsys_opt IS BOUND. "Active BADI Implementation

       CALL BADI lb_sdv03v02_dbsys_opt->xtab_select_sales
         EXPORTING
           ir_smatnr    = s_matnr[]
           ir_swerks    = s_werks[]
           iv_sort_prio = v_sort_prio
           iv_maco_c    = p_maco_c
           iv_ddeliv    = p_ddeliv
           iv_selektion = selektion
           iv_apoactive = gv_apoactive
           iv_logsys    = gv_logsys
         IMPORTING
           et_err_tab   = err_tab
         CHANGING
           ct_matunv    = matunv[]
           ct_xtab      = xtab[].

     ELSEIF cl_db_sys=>is_in_memory_db IS NOT INITIAL.  "e.g. Hana

       CALL METHOD cl_im_sdv03v02_hdb_opt=>if_sdv03v02_dbsys_opt~xtab_select_sales
         EXPORTING
           ir_smatnr    = s_matnr[]
           ir_swerks    = s_werks[]
           iv_sort_prio = v_sort_prio
           iv_maco_c    = p_maco_c
           iv_ddeliv    = p_ddeliv
           iv_selektion = selektion
           iv_apoactive = gv_apoactive
           iv_logsys    = gv_logsys
         IMPORTING
           et_err_tab   = err_tab
         CHANGING
           ct_matunv    = matunv[]
           ct_xtab      = xtab[].

     ELSE.  "Other database

       PERFORM xtab_select_sales USING v_sort_prio.

     ENDIF.

David

[m_olson]

Hi David,

Thanks for chiming in.  No, we are still struggling with the slowness. Our next step is to collect the data we have on our various runs (current PRD vs our HANA QA environment) and open a note with as detailed information as we can about the slowness.

If you ever have a breakthrough would love to hear what your solution was. I plan to update this post with whatever solution we come up with - whenever that may be.

(and yes - thanks for point out that the BADI implementation is moot in the newest version. I hadn't gotten that far when I originally posted this)

Thanks,

Matt