AWR Retention period sizing guidelines Oracle Database Tips by Donald Burleson

For complete details on managing AWR retention and flushing, see Oracle Tuning: The Definitive Reference.

Also see my notes on removing AWR and  disabling AWR data purging.

Out of the box, the AWR automatically captures performance metrics every hour. Some internal database operations, such as the Undo Advisor, use AWR data. The Diagnostics Pack does not have to be licensed to collect the metrics or use this advisor. The Diagnostics Pack will have to be licensed to run reports against the AWR, query any dba_hist_* views, use ASH, leverage ADDM, or use the Performance page in Oracle Enterprise Manager, which will be discussed in the next chapter. Using the Diagnostics Pack in Enterprise Manager is often the biggest time saver.
 
The Automated Workload Repository contains performance metrics gathered once per hour by default and stores the results in the sysaux tablespace. With Oracle RAC databases, the only difference is that the number of instances will linearly scale the AWR data volume. By default, AWR will keep eight days data in the repository. Oracle includes a nice script, $ORACLE_HOME/rdbms/admin/awrinfo.sql, which can help you keep track of the AWR repository sizing.
 
The output of a sample awrinfo.sql script will be discussed in this section. The first portion of the script output shows some introductory information. Notice the snapshot interval and retention. Oddly, this script is warning that there is a non-default AWR setting but this database is out-of-the-box with no custom AWR configuration.
 
~~~~~~~~~~~~~~~
AWR INFO Report
~~~~~~~~~~~~~~~
 
Report generated at 11:44:21 on Sep 16, 2015 ( Tuesday ) in Timezone -06:00                                  
 
Warning: Non Default AWR Setting!                                                                            
----------------------------------------------------------------------------                          
Snapshot interval is 60 minutes and Retention is 8 days                                                      
 
 
       DB_ID DB_NAME HOST_PLATFORM               INST STARTUP_TIME
------------ ------- -------------------------- ----- -----------------
* 2793090278 ORCL    host01 - Linux x86 64-bit      1 10:12:20 (08/28)

Setting an appropriate retention interval for your AWR is critical for proper data retention, especially for predictive modeling.  You can adjust the AWR retention period according to your analysis needs.

In this example the retention period is specified as 3 years (1,576,800 minutes) and the interval between each snapshot is 60 minutes. 

Also see my related notes on dba_hist_wr_control.

DBMS_WORKLOAD_REPOSITORY.create_baseline
(1900, 2000, 'EOY_PEAK_LOADS');

Note:  You must purchase the extra cost Oracle performance pack and Oracle diagnostic packs to use AWR.  Else, you can use the free STATSPACK tables which provide almost the same functionality.

The dbms_workload_repository  package can be used by Oracle DBAs to configure AWR settings such as snapshot interval, flushing and data retention.

The dbms_workload_repository .modify_snapshot_settings procedure is used to configure AWR data retention in the database. The retention interval determines the length of time that data will be preserved in the AWR. The default setting for retention is 10,080 minutes (one week).

desc dbms_workload_repository

PROCEDURE MODIFY_SNAPSHOT_SETTINGS
Argument Name                  Type        In/Out Default?
------------------------------ ----------- ------ -------- RETENTION                      NUMBER      IN     DEFAULT INTERVAL                       NUMBER      IN     DEFAULT DBID                           NUMBER      IN     DEFAULT

The range of valid values for this parameter ranges from ten minutes to 52,560,000 minutes or 100 years.

The dbms_workload_repository package contains the global variables of min_retention and max_retention.  These variables set the lower and upper limits for the retention parameter.

If a zero value is specified for retention, snapshots will be stored for an unlimited amount of time. The following awr_settings.sql script can be used to check the current settings for the AWR interval and retention settings:

awr_settings.sql

select
       extract( day from snap_interval) *24*60+
       extract( hour from snap_interval) *60+
       extract( minute from snap_interval ) "Snapshot Interval",
       extract( day from retention) *24*60+
       extract( hour from retention) *60+
       extract( minute from retention ) "Retention Interval"
from dba_hist_wr_control;

The script returns the current AWR interval values in minutes:

Snapshot Interval Retention Interval
----------------- ------------------
               60              10080

AWR flushing and the MMON background process

The MMON Oracle background process is responsible for periodically flushing the oldest AWR tables, using a LIFO queue method.  Here, we see the flush_level for an AWR installation:

SQL> desc dbms_workload_repository

PROCEDURE CREATE_SNAPSHOT
  Argument Name                  Type                    In/Out Default?
  ------------------------------ ----------------------- ------ --------
  FLUSH_LEVEL                    VARCHAR2                IN     DEFAULT  

FUNCTION CREATE_SNAPSHOT RETURNS NUMBER
  Argument Name                  Type                    In/Out Default?
  ------------------------------ ----------------------- ------ --------
  FLUSH_LEVEL                    VARCHAR2                IN     DEFAULT

The only parameter listed in the procedures is the flush_level , which can have either the default value of TYPICAL or a value of ALL.  When the statistics level is set to ALL, the AWR gathers the maximum amount of performance data.

The MMON background process is responsible for removing old historical data from the AWR. The amount of retention time after which data will be removed from database is determined by the retention setting. However, data can be cleared from the AWR tables by using the dbms_workload_repository.drop_snapshot_range procedure. The starting and ending snapshots for the history to be removed from the AWR will need to be set to run the following script:

desc dbms_workload_repository

PROCEDURE DROP_SNAPSHOT_RANGE
Argument Name                  Type                 In/Out Default?
------------------------------ -------------------- ------ --------
LOW_SNAP_ID                    NUMBER               IN
HIGH_SNAP_ID                   NUMBER               IN
DBID                           NUMBER               IN     DEFAULT

 Disabling and enabling automatic AWR flushing

You can disable the AWR automatic data flushing mechanism at either the system level or disable flushing for individual tables.

For system-wide disable of AWR table flushing you can use these commands but beware that the AWR tablespace will continue to grow ad-infinitum:

alter session set events 'immediate trace name awr_flush_table_off level 99′
alter session set events 'immediate trace name awr_flush_table_off level 106′;

To return to the default of weekly flushing you can issue these commands:

alter session set events 'immediate trace name awr_flush_table_on level 99′;
alter session set events 'immediate trace name awr_flush_table_on level 106′;

If you need to disable flushing the run time statistics for an AWR workload table, you can get the underlying WRH tables with this query:

select
    table_id_kewrtb,
    table_name_kewrtb
from
    x$kewrtb
order by
    table_id_kewrtb;

Once you identify a specific table to disable flushing, you can use an ALTER SYSTEM command:

 alter system set "_awr_disabled_flush_tables"=' WRH$_IC_CLIENT_STATS';