• 1 reply
  • Latest Post - ‏2016-08-16T00:07:45Z by KaranBal
Venkat Podatharapu
Venkat Podatharapu
75 Posts

Pinned topic IBM MDM SE MPI_MEMCMPD.cmpdval changed

‏2016-08-13T21:31:47Z | data management master mdm-migration


We have data loaded in production, during our based load we have cmpdval populated for all the records that is 4 million.

During our daily incremental loads IB process, some how CMPDVAL become paritial meaning stripped off the all demographic information except phone number for few of the members.  so we took these members that we loaded during incremental load ( each of these members processed 5 different times after the base load ) and loaded thru IB in lower environment. 


Daily load Members = 9809809801, 9809809802, 9809809803, 9809809804, 9809809805 with all demographic information brought from production daily loads to replicate in lower enviroment , when we run each day member one time to see which day was the cause of the paritial CMPDVAL, in lower environment it is not repdoucable. All is working without any stripoff the values.

My question is, did any one encountered such kind of issue in your projects, or any one think that this is cause of due to some other configurations, If this is due to MDM Confgiuration project, issue should happen to all the records, but trying to understand what went wrong and causing this.

Due to partitial CMPDVALs for members it is creating Potential Overlay tasks with negative -1.5, -3.0, -4.0  etc during our ongoing daily incremental loads.

If you someone give some input that would be appreciated to my investigation.





  • KaranBal
    227 Posts

    Re: IBM MDM SE MPI_MEMCMPD.cmpdval changed


    We don't lose any data in MDM, and the only reason MDM will delete data in is you have enabled dynamic frequency based bucketing and the bucket exceeds the limit specified; even then it's more buckets than comparison strings. Other reason is if you did a memput in complete mode and only included phone numbers meaning existing data is deleted.


    For anything else, you can enable AlgorithmLog and look into what was happening at the engine level to standardization, comparison strings, and buckets.