Skip to content

Files

Latest commit

 

History

History

faster_stats

First, create a user called STEST.

EDIT user.sql script to use an appropriate tablespace and password (which needs to have about 1.5GB free). EDIT cusr.sql to set the password/connect string

Two 'connect' scripts are available: cadm.sql <-- Connects to SYSDBA account cusr.sql <-- Connects to the STEST user account - EDIT THIS TO SET YOUR PASSWORD You can edit these to suit you system, particularly if you are using a multitenant environment.

Create tables: @tabs.sql

Gather database stats to start with a 'clean slate': @gather.sql

Make RM plan: @make_plan.sql Note that SCOPE=MEMORY so the plan change is temporary. If necessary, you can drop the plan with drop_plan.sql

If you want to use the DEFAULT_MAINTENANCE_PLAN instead, run modify_plan.sql in the 'DEFAULT' directory.

To use AUTO_DEGREE: @auto.sql To use serial execution: @noauto.sql

To user CONCURRENT: @conc.sql To disable CONCURRENT: @noconc.sql

Prepare the test by dripping stats on the test tables: @drop_stats.sql

Initiate stats gathering by creating a 20min batch window: @run_gather.sql If you want to use DEFAULT_MAINTENANCE_PLAN, then run 'run_gather.sql' in the 'DEFAULT' subdirectory instead.

I have put the monitoring queries I use in util.sql Take a look at scheduler information: get_status.sql Take a look at stale objects: stale.sql

If you want to stop the test: drop_window.sql [stop.sql is an alternative]

========================

You can run a manual stats gathering with 'manual.sql'