Table of Contents

Search

  1. Introduction
  2. Configuring Hub Console Tools
  3. Building the Data Model
  4. Configuring the Data Flow
  5. Executing Informatica MDM Hub Processes
  6. Configuring Application Access
  7. MDM Hub Properties
  8. Viewing Configuration Details
  9. Search with Solr
  10. Row-level Locking
  11. MDM Hub Logging
  12. Table Partitioning
  13. Collecting MDM Environment Information with the Product Usage Toolkit
  14. Glossary

Limiting the Parallel Degree for Gathering Statistics in Oracle environments

Limiting the Parallel Degree for Gathering Statistics in Oracle environments

The parallel degree for gathering statistics in Oracle environments should be limited to ensure Hub process performance is not negatively impacted.
You must be logged in as a user with database administrator privileges to set the parallel degree.
Perform the following steps to assign an appropriate parallel degree for gathering statistics:
  1. Calculate the appropriate parallel degree by using the following formula:
    APPROPRIATE PARALLEL DEGREE = CPU_COUNT * PARALLEL_THREADS_PER_CPU
    CPU_COUNT
    is the number of CPUs available for Oracle to use. The
    PARALLEL_THREADS_PER_CPU
    is usually
    2
    .
    If the server has many CPUs, select a parallel degree value equal to or less than the number of CPU cores.
    If other applications are running on the same server as the Hub, decide how many CPU resources can be allocated to the Hub and then use this number when setting the appropriate parallel degree.
  2. Check the current parallel degree setting by running the following SQL*Plus command:
    select DBMS_STATS.GET_PREFS( 'DEGREE' ) from dual;
  3. If necessary, set the appropriate parallel degree by running one of the following SQL*Plus commands:
    • In Oracle 10g:
      DBMS_STATS.SET_PARAM ('DEGREE',<
      appropriate parallel degree value
      >);
    • In Oracle 11g:
      DBMS_STATS.SET_GLOBAL_PREFS ('DEGREE', <
      appropriate parallel degree value
      >);
  4. Test performance using the new parallel degree value by running the following SQL command for a large table:
    DBMS_STATS.GATHER_TABLE_STATS
  5. Repeat steps 3 and 4, reducing the parallel degree each time, until wait events are eliminated and performance is acceptable.

0 COMMENTS

We’d like to hear from you!