Partition processing peak RRS feed

  • Question

  • Hi

    I have a S0 10GB service running, I recently upgraded from basic to be able to do partition processing since we kept hitting the upper memory limit during full processing. 

    I now do a full process on the 2020 partition, which contains about 4-5% of the data (total 2.5-3GB when idle).

    But the 2020 partition processing peak is almost similar to processing the entire model without partitions.

    Am I missing something?

    My next step is clearing the partition, wait a couple of minutes, and then full process 2020, but it seems wrong.

    Thanks in advance

    Best regards


    Monday, February 3, 2020 11:46 AM

All replies

  • Hello BRNDK, 

    Apologizes for the late response on this . 

    I will reach out to internal teams for more information.

    Thanks Himanshu

    Tuesday, February 4, 2020 5:52 PM
  • Hello ,

    Are you having many calculated columns\calculated tables here ?.

    Thanks Himanshu

    Wednesday, February 5, 2020 7:29 AM
  • Hi Himanshu

    Thanks for getting back.

    I have 1 table, 16 regular colums (mostly whole number references to dimension tables) and 5 calculated columns (3 whole numbers and 2 booleans).

    Maybe all the tables calculated columns re-evaluated even though only one partition is updated?
    (In that case, I think I could move some of the logic into the ETL proces)

    Best regards


    Wednesday, February 5, 2020 6:41 PM
  • That does not sound to be too many . I suggest you to please move the logic to ETL ( if you can ) and see if that help . At this time I am not very confident . Please do let us know how it goes .

    Thanks Himanshu

    Wednesday, February 5, 2020 10:13 PM
  • I will do that as soon as possible, since I will eventually hit the barrier again.

    Do You suspect that all partitions' calculated fields are re-evaluated when only one partition is updated?

    Thanks for getting back :)

    Best regards


    Thursday, February 6, 2020 7:17 AM