We are using Pega 8.3.6 and facing a huge problem in data flow, in backend pega is running below query for each of the partition for each queue processor

UPDATE PEGADATA.pr_data_decision_df_part "PC0" SET "PXLASTIDS" = :1

         ,"PXRECORDSPROCESSED" = :2  ,"PXUPDATEDATETIME" = :3   WHERE  (

         "PC0"."PXWORKID" = :4   AND "PC0"."PXVALUES" = :5   )

Because of that we are having concurrency lock. 

Is there any way in 8.3.6 we can reduce the partition count for each queue processor? then how to reduce the whole kafka partition count without having any data loss.