Db2

Db2

Connect with Db2, Informix, Netezza, open source, and other data experts to gain value from your data, share insights, and solve problems.

 View Only

Testresults Achmea DB2 AIX 11.1 in the EAP program (upload2/2)

  • 1.  Testresults Achmea DB2 AIX 11.1 in the EAP program (upload2/2)

    Posted Wed March 31, 2021 12:38 PM

    Dear member,

     

    I've tested some queries with release DB2 11.1 of the EAP progam.

    We upload the testresults in tree uploads: (upload 1/2, upload2/2, upload 2/2a)

    We had the following main issues with release 10.5.5.

     

       - problems with queries (joins) on large tables regarding the sort heap memory pools.

     

       - the absence of creating IDENTITY columns in column organized tables

     

     

    The global results of the tests :

     

        - column organized tables with IDENTITY-columns are possible in DB2 11.1 and all functions, including

     

          "restart ID"  are working well.

     

        - the sort memory problems still exists.

     

          I've tested 4 queries, that caused problems in 10.5 with sortheap memory

     

         The same problems (memory overflow on the sortheap memory) occurred in DB2 11.1.

     

          The four queries are tested each with three different sortheap parameters (1.6 Gb, 3.2 Gb and 6.4 GB),

     

          all with a sortheap shared parameter of 32 Gb.

     

          Finally all the tests are running well with a sortheap of 6.4 Gb, but then we can only execute 5 (sort)-queries

     

          in parallel. and no guarantees that other queries are running with that parameter !!

     

     

    Finally our "wish-list" for column organized tables :

     

        - online schema changes :

     

            - only "alter table add column" clause possible

     

            - no datatype changes, etc.  (as in DB2 10.5) 

     

        - no LONG datatypes (BLOB, CLOB and LONG VARCHAR)

     

        - no PARTITIONED tables   (query data-independence, detach/attach partitions)

     

        - no classic reorg (clustering/ordering the synopsys table ??)

     

     

    Below a schema of the tested queries, they all ran with a fresh activated database.

     

    These are the four queries, each tested with three different sortheap values and all with a sortheap shared of 32 Gb.

     

     

    QUERY

    ENDED

    MAX_SORT_CONSUMER_SORTHEAP_TOP

    MAX_SORT_MEMORY_USED

    MAX_SORT_SHRHEAP

    SORTHEAP

    Q11

    ABEND SORT_MEMORY

    1600

    2798

    2637

    1600

     

    Q12

    ABEND SORT_MEMORY

    3200

    4701

    5046

    3200

     

    Q13

    SUCCEEDED

    4483

    6246

    6400

    6400

     

    Q21

    SUCCEEDED

    1600

    4812

    5118

    1600

     

    Q22

    SUCCEEDED

    3200

    7851

    9906

    3200

     

    Q23

    SUCCEEDED

    6400

    12483

    19574

    6400

     

    Q31

    ABEND SORT_MEMORY

    1600

    4641

    5438

    1600

     

    Q32

    ABEND SORT_MEMORY

    3200

    8038

    7596

    3200

     

    Q33

    SUCCEEDED

    4589

    9445

    8964

    6400

     

    Q41

    SUCCEEDED

    1600

    7645

    4401

    1600

     

    Q42

    SUCCEEDED

    533

    750

    3200

    3200

     

    Q43

    SUCCEEDED

    1066

    746

    6400

    6400

     
                 
                             

     

     

    During the tests I've inserted each 2 or 3 seconds a record in a table with the actual sort-parameter values.

     

    The results of the 12 tests, including db2diag.log messages in case of abends, are saved and the actual sort values

     

    during the tests are saved in an Excell sheet.

     

    I'll send all the results in appendixes

     

     

    Best regards,

     

    Jan Bouwmeester

     

    DBA Achmea The Netherlands


    #Db2
    #Db2EarlyAccessProgram(EAP)Forum