Apex Schema Optimizer Statistics

Providing the Oracle optimizer with accurate statistics on the Apex schema tables can be challenging.

Many of tables tracking Apex session state information see a huge turnover in data such that statistics gathered overnight can quickly become stale resulting in poor choice of execution plan by the optimizer and seriously degraded performance.

There are several ways to tackle the issue e.g. locking the stats on some of the tables to values that are known to perform reliably.

However I believe it’s better to give the optimizer the most accurate information you can, so it can make more informed decisions based on the actual data in the tables.

The problem is that even if you refresh the optimizer statistics often, you need to ensure that an execution plans that are stored in the shared_pool are also cleared or you risk still using a plan that does not reflect those new statistics.

Fortunately you can achieve all this with one relatively straight forward job i.e.

conn / as sysdba
variable v_jobno number;
'dbms_stats.gather_schema_stats(ownname=>''APEX_040200'',options=>''GATHER STALE'',cascade=>true,no_invalidate=>false);',

This will create a job that runs every five minutes and refresh statistics on any Apex tables that are marked as stale, but also thanks to the no_invalidate option, it will clear any plans from the shared_pool that would be affected by the new statistics gathered.

You should make sure the schema reflects your Apex version, but with this job in place you should have no worries about stale statistics on the Apex session state tables impacting your application performance.