 |
|
Oracle 11g extended optimizer
statistics
Oracle11g Tips by Burleson Consulting
December 13, 2007 |
One of the most exciting new features
of Oracle 11g is improvements to the dbms_stats package,
specifically the ability to aid complex queries by providing
extended statistics to the cost-based optimizer (CBO).
The 11g extended optimizer
statistics are intended to improve the optimizers guesses for the
cardinality of combined columns and columns that are modified by a
built-in or user-defined function.
In Oracle 10g we see that dynamic
sampling can be used to provide inter-table cardinality estimates,
but dynamic sampling has important limitations. However, the
11g extended statistics in dbms_stats relieves much of the
problem of sub-optimal table join orders.
In the absence of column histograms
and extended statistics, the Oracle cost-based optimizer must be
able to “guess” the size of complex result sets information, and it
sometimes gets it wrong. This is one reason why the ORDERED hint is
one of the most popular SQL tuning hints; using the ORDERED hint
allows you to specify that the tables be joined together in the same
order that they appear in the FROM clause.
In this example, the four-way table
join only returns 18 rows, but the query carries 9,000 rows in
intermediate result sets, slowing-down the SQL execution speed:
A suboptimal table join order
If we
were able to predict the sizes of the intermediate results, we can
re-sequence the table-join order to carry less “intermediate
baggage” during the four-way table join, in this example carrying
only 3,000 intermediate rows between the table joins:
11g extended statistics help
the CBO predict inter-table join result set sizes
Let's take a closer look and understand how the
11g extended dbms_stats data helps the optimizer make better
guesses of result set sizes.
Inside extended optimizer statistics
The new 11g dbms_stats package has
several new procedures to aid in supplementing histogram data, and
the state of these extended histograms can be seen in the
user_tab_col_statistics view:
-
dbms_stats.create_extended_stats
-
dbms_stats.show_extended_stats_name
-
dbms_stats.drop_extended_stats
Arup Nanda has a great article on extended statistics with
dbms_stats, specialty histogram analysis using function-based
columnar data:
Next, re-gather statistics on the table and
collect the extended statistics on the expression upper(cust_name).
begin
dbms_stats.gather_table_stats (
ownname => 'ARUP',
tabname => 'CUSTOMERS',
method_opt => 'for all columns size skewonly for columns (upper(cust_name))'
);
end;
Alternatively you
can define the column group as part of the gather statistics
command.
You do that by
placing these columns in the method_opt parameter of the
gather_table_stats procedure in dbms_stats as shown below:
begin
dbms_stats.gather_table_stats (
ownname => 'ARUP',
tabname => 'BOOKINGS',
estimate_percent=> 100,
method_opt => 'FOR ALL COLUMNS SIZE SKEWONLY FOR COLUMNS(HOTEL_ID,RATE_CATEGORY)',
cascade => true
This is a work in
progress excerpt from
the book "Oracle
11g New Features" by Rampant TechPress.
See these related
notes on using 11g CBO statistics:
 |
If you like Oracle tuning, you may enjoy my new book "Oracle
Tuning: The Definitive Reference", over 900 pages
of BC's favorite tuning tips & scripts.
You can buy it direct from the publisher for 30%-off and get
instant access to the code depot of Oracle tuning scripts. |
 |
|
Need an Oracle Health Check?
- Do you have
bad performance after an upgrade?
- Need to
certify that your database follows best practices?
BC Oracle performance gurus can quickly
certify every aspect of your
Oracle database and provide a complete verification that your database
is fully optimized. |
|
|
