19 Using Optimizer Hints

19.1.1 Types of Hints

Hints can be of the following general types:

• Single-table

  Single-table hints are specified on one table or view. INDEX and USE_NL are examples of single-table hints.

• Multi-table

  Multi-table hints are like single-table hints, except that the hint can specify one or more tables or views. LEADING is an example of a multi-table hint. Note that USE_NL(table1 table2) is not considered a multi-table hint because it is actually a shortcut for USE_NL(table1) and USE_NL(table2).

• Query block

  Query block hints operate on single query blocks. STAR_TRANSFORMATION and UNNEST are examples of query block hints.

• Statement

  Statement hints apply to the entire SQL statement. ALL_ROWS is an example of a statement hint.

19.1.2 Hints by Category

Optimizer hints are grouped into the following categories:

• Hints for Optimization Approaches and Goals

• Hints for Access Paths

• Hints for Query Transformations

• Hints for Join Orders

• Hints for Join Operations

• Hints for Parallel Execution

• Additional Hints

These categories, and the hints contained within each category, are listed in the sections that follow.

19.2.1 Specifying a Full Set of Hints

When using hints, in some cases, you might need to specify a full set of hints in order to ensure the optimal execution plan. For example, if you have a very complex query, which consists of many table joins, and if you specify only the INDEX hint for a given table, then the optimizer needs to determine the remaining access paths to be used, as well as the corresponding join methods. Therefore, even though you gave the INDEX hint, the optimizer might not necessarily use that hint, because the optimizer might have determined that the requested index cannot be used due to the join methods and access paths selected by the optimizer.

In Example 19-1, the LEADING hint specifies the exact join order to be used; the join methods to be used on the different tables are also specified.

SELECT /*+ LEADING(e2 e1) USE_NL(e1) INDEX(e1 emp_emp_id_pk) 
           USE_MERGE(j) FULL(j) */
    e1.first_name, e1.last_name, j.job_id, sum(e2.salary) total_sal
  FROM employees e1, employees e2, job_history j
  WHERE e1.employee_id = e2.manager_id
    AND e1.employee_id = j.employee_id
    AND e1.hire_date = j.start_date
  GROUP BY e1.first_name, e1.last_name, j.job_id
  ORDER BY total_sal;

19.2.2 Specifying a Query Block in a Hint

To identify a query block in a query, an optional query block name can be used in a hint to specify the query block to which the hint applies. The syntax of the query block argument is of the form @queryblock, where queryblock is an identifier that specifies a query block in the query. The queryblock identifier can either be system-generated or user-specified.

• The system-generated identifier can be obtained by using EXPLAIN PLAN for the query. Pre-transformation query block names can be determined by running EXPLAIN PLAN for the query using the NO_QUERY_TRANSFORMATION hint.

• The user-specified name can be set with the QB_NAME hint.

In Example 19-2, the query block name is used with the NO_UNNEST hint to specify a query block in a SELECT statement on the view.

CREATE OR REPLACE VIEW v AS
  SELECT e1.first_name, e1.last_name, j.job_id, sum(e2.salary) total_sal
    FROM employees e1,
      ( SELECT *
          FROM employees e3) e2, job_history j
         WHERE e1.employee_id = e2.manager_id
          AND e1.employee_id = j.employee_id
          AND e1.hire_date = j.start_date
          AND e1.salary = ( SELECT max(e2.salary) 
                             FROM employees e2 
                             WHERE e2.department_id = e1.department_id ) 
  GROUP BY e1.first_name, e1.last_name, j.job_id
  ORDER BY total_sal;

SELECT PLAN_TABLE_OUTPUT FROM TABLE(DBMS_XPLAN.DISPLAY(NULL, NULL, 'SERIAL'));
...
Query Block Name / Object Alias (identified by operation id):
-------------------------------------------------------------
...
  10 - SEL$4        / E2@SEL$4

SELECT /*+ NO_UNNEST( @SEL$4 ) */
  *
 FROM v;

19.2.3 Specifying Global Table Hints

Hints that specify a table generally refer to tables in the DELETE, SELECT, or UPDATE query block in which the hint occurs, not to tables inside any views referenced by the statement. When you want to specify hints for tables that appear inside views, Oracle recommends using global hints instead of embedding the hint in the view. Table hints described in this chapter can be transformed into a global hint by using an extended tablespec syntax that includes view names with the table name.

In addition, an optional query block name can precede the tablespec syntax. See "Specifying a Query Block in a Hint".

Hints that specify a table use the following syntax:

tablespec::=
Description of the illustration tablespec.gif

where:

• view specifies a view name

• table specifies the name or alias of the table

If the view path is specified, the hint is resolved from left to right, where the first view must be present in the FROM clause, and each subsequent view must be specified in the FROM clause of the preceding view.

For example, in Example 19-3 a view v is created to return the first and last name of the employee, his or her first job and the total salary of all direct reports of that employee for each employee with the highest salary in his or her department. When querying the data, you want to force the use of the index emp_job_ix for the table e3 in view e2.

CREATE OR REPLACE VIEW v AS
SELECT
    e1.first_name, e1.last_name, j.job_id, sum(e2.salary) total_sal
FROM employees e1,
        ( SELECT *
          FROM employees e3) e2, job_history j
WHERE e1.employee_id = e2.manager_id
     AND e1.employee_id = j.employee_id
     AND e1.hire_date = j.start_date
     AND e1.salary = ( SELECT
                       max(e2.salary)
                       FROM employees e2
                       WHERE e2.department_id = e1.department_id)
GROUP BY e1.first_name, e1.last_name, j.job_id
ORDER BY total_sal;

SELECT /*+ INDEX(v.e2.e3 emp_job_ix) */  * 
  FROM v;

SELECT /*+ INDEX(@SEL$2 e2.e3 emp_job_ix) */ * 
  FROM v;

SELECT /*+ INDEX(@SEL$3 e3 emp_job_ix) */ * 
  FROM v;

CREATE OR REPLACE VIEW v1 AS
  SELECT *
    FROM employees
    WHERE employee_id < 150;

CREATE OR REPLACE VIEW v2 AS
  SELECT v1.employee_id employee_id, departments.department_id department_id
    FROM v1, departments
    WHERE v1.department_id = departments.department_id;

SELECT /*+ NO_MERGE(v2) INDEX(v2.v1.employees emp_emp_id_pk)
                        FULL(v2.departments) */ *
  FROM v2
  WHERE department_id = 30;

19.2.4 Specifying Complex Index Hints

Hints that specify an index can use either a simple index name or a parenthesized list of columns as follows:

indexspec::=

where:

• table specifies the name

• column specifies the name of a column in the specified table

  • The columns can optionally be prefixed with table qualifiers allowing the hint to specify bitmap join indexes where the index columns are on a different table than the indexed table. If tables qualifiers are present, they must be base tables, not aliases in the query.

  • Each column in an index specification must be a base column in the specified table, not an expression. Function-based indexes cannot be hinted using a column specification unless the columns specified in the index specification form the prefix of a function-based index.

• index specifies an index name

  When tablespec is followed by indexspec in the specification of a hint, a comma separating the table name and index name is permitted but not required. Commas are also permitted, but not required, to separate multiple occurrences of indexspec.

The hint is resolved as follows:

• If an index name is specified, only that index is considered.

• If a column list is specified and an index exists whose columns match the specified columns in number and order, only that index is considered. If no such index exists, then any index on the table with the specified columns as the prefix in the order specified is considered. In either case, the behavior is exactly as if the user had specified the same hint individually on all the matching indexes.

For example, in Example 19-3 the job_history table has a single-column index on the employee_id column and a concatenated index on employee_id and start_date columns. To specifically instruct the optimizer on index use, the query can be hinted as follows:

SELECT /*+ INDEX(v.j jhist_employee_ix (employee_id start_date)) */ * FROM v;

19.3.1 Hints and Complex Views

By default, hints do not propagate inside a complex view. For example, if you specify a hint in a query that selects against a complex view, then that hint is not honored, because it is not pushed inside the view.

Unless the hints are inside the base view, they might not be honored from a query against the view.

19.3.2 Hints and Mergeable Views

This section describes hint behavior with mergeable views.

Optimization Approaches and Goal Hints in Views

Optimization approach and goal hints can occur in a top-level query or inside views.

• If there is such a hint in the top-level query, then that hint is used regardless of any such hints inside the views.

• If there is no top-level optimizer mode hint, then mode hints in referenced views are used as long as all mode hints in the views are consistent.

• If two or more mode hints in the referenced views conflict, then all mode hints in the views are discarded and the session mode is used, whether default or user-specified.

Access Path and Join Hints on Views

Access path and join hints on referenced views are ignored, unless the view contains a single table (or references an Additional Hints view with a single table). For such single-table views, an access path hint or a join hint on the view applies to the table inside the view.

Access Path and Join Hints Inside Views

Access path and join hints can appear in a view definition.

• If the view is an inline view (that is, if it appears in the FROM clause of a SELECT statement), then all access path and join hints inside the view are preserved when the view is merged with the top-level query.

• For views that are non-inline views, access path and join hints in the view are preserved only if the referencing query references no other tables or views (that is, if the FROM clause of the SELECT statement contains only the view).

Parallel Execution Hints on Views

PARALLEL, NO_PARALLEL, PARALLEL_INDEX, and NO_PARALLEL_INDEX hints on views are applied recursively to all the tables in the referenced view. Parallel execution hints in a top-level query override such hints inside a referenced view.

Parallel Execution Hints Inside Views

PARALLEL, NO_PARALLEL, PARALLEL_INDEX, and NO_PARALLEL_INDEX hints inside views are preserved when the view is merged with the top-level query. Parallel execution hints on the view in a top-level query override such hints inside a referenced view.

19.3.3 Hints and Nonmergeable Views

With nonmergeable views, optimization approach and goal hints inside the view are ignored; the top-level query decides the optimization mode.

Because nonmergeable views are optimized separately from the top-level query, access path and join hints inside the view are preserved. For the same reason, access path hints on the view in the top-level query are ignored.

However, join hints on the view in the top-level query are preserved because, in this case, a nonmergeable view is similar to a table.