Oracle Logbook

March 9, 2012

Reducing redo generation

Filed under: Uncategorized — Helio Dias @ 8:06 pm

Log less, a dream feature for many DBAs, but Log is a matter of how much you write, so this article is gonna
show workarounds to reduce this issue.

When one have a small archive area , one solution might be decrease the redo log activity, therefore minimizing

the buffer written activity.
To accomplish it, I´d suggest:

Evaluating the possibility to split the developer data into another instance non archived (if such case).

Dropping not helpful indexes.
Indexes write generate more redo log than the table write indeed.

Changing ordinary tables used as temporary table, to global temporary table.
Global temporary table almost don´t write redo.

Dropping and recreating indexes before data load operation.

Check the frequency of refresh on materialized views.
Less refresh = less redo.

Export and import tables with lower PCT Free.
Less blocks, means less redo generation.

Rebuild Indexes, following the above principle.

Recreate some tables as IOTs
Less objects (Table + PK) less redo.
Increase the cache of the sequences.
Every time the database have to get a new set of values it write a redo log

 

 

Blog at WordPress.com.