Query Hints

http://msdn.microsoft.com/en-us/library/ms181714.aspx 

Syntax

<query_hint > ::= 

{ { HASH | ORDER } GROUP 

  | { CONCAT | HASH | MERGE } UNION 

  | { LOOP | MERGE | HASH } JOIN 

  | FAST number_rows 

  | FORCE ORDER 

  | MAXDOP number_of_processors 

  | OPTIMIZE FOR ( @variable_name { UNKNOWN | = literal_constant } [ , ...n ] )

  | OPTIMIZE FOR UNKNOWN

  | PARAMETERIZATION { SIMPLE | FORCED }

  | RECOMPILE

  | ROBUST PLAN 

  | KEEP PLAN 

  | KEEPFIXED PLAN

  | EXPAND VIEWS 

  | MAXRECURSION number 

  | USE PLAN N'xml_plan'

    | TABLE HINT ( exposed_object_name  [ , <table_hint> [ [, ]...n ] ] )

<table_hint> ::=

[ NOEXPAND ] { 

    INDEX ( index_value [ ,...n ] ) | INDEX = ( index_value )

  | FASTFIRSTROW 

  | FORCESEEK

  | HOLDLOCK 

  | NOLOCK 

  | NOWAIT

  | PAGLOCK 

  | READCOMMITTED 

  | READCOMMITTEDLOCK 

  | READPAST 

  | READUNCOMMITTED 

  | REPEATABLEREAD 

  | ROWLOCK 

  | SERIALIZABLE 

  | TABLOCK 

  | TABLOCKX 

  | UPDLOCK 

  | XLOCK 

}

 Arguments

{ HASH | ORDER } GROUP

Specifies that aggregations described in the GROUP BY, DISTINCT, or COMPUTE clause of the query should use hashing or ordering.

{ MERGE | HASH | CONCAT } UNION

Specifies that all UNION operations are performed by merging, hashing, or concatenating UNION sets. If more than one UNION hint is specified, the query optimizer selects the least expensive strategy from those hints specified.

{ LOOP | MERGE | HASH } JOIN

Specifies that all join operations are performed by LOOP JOIN, MERGE JOIN, or HASH JOIN in the whole query. If more than one join hint is specified, the optimizer selects the least expensive join strategy from the allowed ones.

If, in the same query, a join hint is also specified in the FROM clause for a specific pair of tables, this join hint takes precedence in the joining of the two tables, although the query hints still must be honored. Therefore, the join hint for the pair of tables may only restrict the selection of allowed join methods in the query hint. For more information, see Join Hints (Transact-SQL).

FAST number_rows

Specifies that the query is optimized for fast retrieval of the first number_rows. This is a nonnegative integer. After the first number_rows are returned, the query continues execution and produces its full result set.

FORCE ORDER

Specifies that the join order indicated by the query syntax is preserved during query optimization. Using FORCE ORDER does not affect possible role reversal behavior of the query optimizer. For more information, see Understanding Hash Joins.

Note:
In a MERGE statement, the source table is accessed before the target table as the default join order, unless the WHEN SOURCE NOT MATCHED clause is specified. Specifying FORCE ORDER preserves this default behavior.

For information about how the SQL Server query optimizer enforces the FORCE ORDER hint when a query contains a view, see View Resolution.

MAXDOP number

Overrides the max degree of parallelism configuration option of sp_configure and Resource Governor for the query specifying this option. The MAXDOP query hint can exceed the value configured with sp_configure. If MAXDOP exceeds the value configured with Resource Governor, the Database Engine uses the Resource Governor MAXDOP value, described in ALTER WORKLOAD GROUP (Transact-SQL). All semantic rules used with the max degree of parallelism configuration option are applicable when you use the MAXDOP query hint. For more information, see max degree of parallelism Option.

OPTIMIZE FOR ( @variable_name { UNKNOWN | = literal_constant } [ , ... n ] )

Instructs the query optimizer to use a particular value for a local variable when the query is compiled and optimized. The value is used only during query optimization, and not during query execution.

@variable_name

Is the name of a local variable used in a query, to which a value may be assigned for use with the OPTIMIZE FOR query hint.

UNKNOWN

Specifies that the query optimizer use statistical data instead of the initial value to determine the value for a local variable during query optimization.

literal_constant

Is a literal constant value to be assigned @variable_name for use with the OPTIMIZE FOR query hint. literal_constant is used only during query optimization, and not as the value of @variable_name during query execution. literal_constant can be of any SQL Server system data type that can be expressed as a literal constant. The data type of literal_constant must be implicitly convertible to the data type that @variable_name references in the query.

OPTIMIZE FOR can counteract the default parameter detection behavior of the optimizer or can be used when you create plan guides. For more information, see Recompiling Stored Procedures and Optimizing Queries in Deployed Applications by Using Plan Guides.

OPTIMIZE FOR UNKNOWN

Instructs the query optimizer to use statistical data instead of the initial values for all local variables when the query is compiled and optimized, including parameters created with forced parameterization. For more information about forced parameterization, see Forced Parameterization.

If OPTIMIZE FOR @variable_name = literal_constant and OPTIMIZE FOR UNKNOWN are used in the same query hint, the query optimizer will use the literal_constant that is specified for a specific value and UNKNOWN for the remaining variable values. The values are used only during query optimization, and not during query execution.

PARAMETERIZATION { SIMPLE | FORCED }

Specifies the parameterization rules that the SQL Server query optimizer applies to the query when it is compiled.

Important:
The PARAMETERIZATION query hint can only be specified inside a plan guide. It cannot be specified directly within a query.

SIMPLE instructs the query optimizer to attempt simple parameterization. FORCED instructs the optimizer to attempt forced parameterization. The PARAMETERIZATION query hint is used to override the current setting of the PARAMETERIZATION database SET option inside a plan guide. For more information, see Specifying Query Parameterization Behavior by Using Plan Guides.

RECOMPILE

Instructs the SQL Server Database Engine to discard the plan generated for the query after it executes, forcing the query optimizer to recompile a query plan the next time the same query is executed. Without specifying RECOMPILE, the Database Engine caches query plans and reuses them. When compiling query plans, the RECOMPILE query hint uses the current values of any local variables in the query and, if the query is inside a stored procedure, the current values passed to any parameters.

RECOMPILE is a useful alternative to creating a stored procedure that uses the WITH RECOMPILE clause when only a subset of queries inside the stored procedure, instead of the whole stored procedure, must be recompiled. For more information, see Recompiling Stored Procedures. RECOMPILE is also useful when you create plan guides. For more information, see Optimizing Queries in Deployed Applications by Using Plan Guides.

ROBUST PLAN

Forces the query optimizer to try a plan that works for the maximum potential row size, possibly at the expense of performance. When the query is processed, intermediate tables and operators may have to store and process rows that are wider than any one of the input rows. The rows may be so wide that, sometimes, the particular operator cannot process the row. If this occurs, the Database Engine produces an error during query execution. By using ROBUST PLAN, you instruct the query optimizer not to consider any query plans that may encounter this problem.

If such a plan is not possible, the query optimizer returns an error instead of deferring error detection to query execution. Rows may contain variable-length columns; the Database Engine allows for rows to be defined that have a maximum potential size beyond the ability of the Database Engine to process them. Generally, despite the maximum potential size, an application stores rows that have actual sizes within the limits that the Database Engine can process. If the Database Engine encounters a row that is too long, an execution error is returned.

KEEP PLAN

Forces the query optimizer to relax the estimated recompile threshold for a query. The estimated recompile threshold is the point at which a query is automatically recompiled when the estimated number of indexed column changes have been made to a table by running UPDATE, DELETE, MERGE, or INSERT statements. Specifying KEEP PLAN makes sure that a query will not be recompiled as frequently when there are multiple updates to a table.

KEEPFIXED PLAN

Forces the query optimizer not to recompile a query due to changes in statistics. Specifying KEEPFIXED PLAN makes sure that a query will be recompiled only if the schema of the underlying tables is changed or if sp_recompile is executed against those tables.

EXPAND VIEWS

Specifies that the indexed views are expanded and the query optimizer will not consider any indexed view as a substitute for any part of the query. A view is expanded when the view name is replaced by the view definition in the query text.

This query hint virtually disallows direct use of indexed views and indexes on indexed views in the query plan.

The indexed view is not expanded only if the view is directly referenced in the SELECT part of the query and WITH (NOEXPAND) or WITH (NOEXPAND, INDEX( index_value [ ,...n ] ) ) is specified. For more information about the query hint WITH (NOEXPAND), see FROM.

Only the views in the SELECT part of statements, including those in INSERT, UPDATE, MERGE, and DELETE statements are affected by the hint.

MAXRECURSION number

Specifies the maximum number of recursions allowed for this query. number is a nonnegative integer between 0 and 32767. When 0 is specified, no limit is applied. If this option is not specified, the default limit for the server is 100.

When the specified or default number for MAXRECURSION limit is reached during query execution, the query is ended and an error is returned.

Because of this error, all effects of the statement are rolled back. If the statement is a SELECT statement, partial results or no results may be returned. Any partial results returned may not include all rows on recursion levels beyond the specified maximum recursion level.

For more information, see WITH common_table_expression (Transact-SQL).

USE PLAN N ' xml_plan '

Forces the query optimizer to use an existing query plan for a query that is specified by 'xml_plan'. For more information, see Specifying Query Plans with Plan Forcing. USE PLAN cannot be specified with INSERT, UPDATE, MERGE, or DELETE statements.

TABLE HINT ( exposed_object_name [ , <table_hint> [ [ , ]... n ] ] )

Applies the specified table hint to the table or view that corresponds to exposed_object_name. We recommend using a table hint as a query hint only in the context of a plan guide.

exposed_object_name can be one of the following references:

• When an alias is used for the table or view in the FROM clause of the query, exposed_object_name is the alias.
  
• When an alias is not used, exposed_object_name is the exact match of the table or view referenced in the FROM clause. For example, if the table or view is referenced using a two-part name, exposed_object_name is the same two-part name.
  

When exposed_object_name is specified without also specifying a table hint, any indexes specified in the query as part of a table hint for the object are disregarded and index usage is determined by the query optimizer. You can use this technique to eliminate the effect of an INDEX table hint when you cannot modify the original query. See Example J.

<table_hint> ::= { [ NOEXPAND ] { INDEX ( index_value [ ,... n ] ) | INDEX = ( index_value ) | FASTFIRSTROW | FORCESEEK | HOLDLOCK | NOLOCK | NOWAIT | PAGLOCK | READCOMMITTED | READCOMMITTEDLOCK | READPAST | READUNCOMMITTED | REPEATABLEREAD | ROWLOCK | SERIALIZABLE | TABLOCK | TABLOCKX | UPDLOCK | XLOCK }

Is the table hint to apply to the table or view that corresponds to exposed_object_name as a query hint. For a description of these hints, see Table Hints (Transact-SQL).

Table hints other than INDEX and FORCESEEK are disallowed as query hints unless the query already has a WITH clause specifying the table hint. For more information, see Remarks.

 Remarks

Query hints cannot be specified in an INSERT statement except when a SELECT clause is used inside the statement.

Query hints can be specified only in the top-level query, not in subqueries. When a table hint is specified as a query hint, the hint can be specified in the top-level query or in a subquery; however, the value specified for exposed_object_name in the TABLE HINT clause must match exactly the exposed name in the query or subquery.

Specifying Table Hints as Query Hints

We recommend using the INDEX or FORCESEEK table hint as a query hint only in the context of a plan guide. Plan guides are useful when you cannot modify the original query, for example, because it is a third-party application. The query hint specified in the plan guide is added to the query before it is compiled and optimized. For ad-hoc queries, use the TABLE HINT clause only when testing plan guide statements. For all other ad-hoc queries, we recommend specifying these hints only as table hints.

When specified as a query hint, the INDEX and FORCESEEK table hints are valid for the following objects:

• Tables
  
• Views
  
• Indexed views
  
• Common table expressions (The hint must be specified in the SELECT statement whose result set populates the common table expression.)
  
• Dynamic management views
  
• Named subqueries
  

The INDEX and FORCESEEK table hints can be specified as query hints for a query that does not have any existing table hints, or they can be used to replace existing INDEX or FORCESEEK hints in the query, respectively. Table hints other than INDEX and FORCESEEK are disallowed as query hints unless the query already has a WITH clause specifying the table hint. In this case, a matching hint must also be specified as a query hint by using TABLE HINT in the OPTION clause to preserve the semantics of the query. For example, if the query contains the table hint NOLOCK, the OPTION clause in the @hints parameter of the plan guide must also contain the NOLOCK hint. See Example K. When a table hint other than INDEX or FORCESEEK is specified by using TABLE HINT in the OPTION clause without a matching query hint, or vice versa; error 8702 is raised (indicating that the OPTION clause can cause the semantics of the query to change) and the query fails. For more information, see Using the INDEX and FORCESEEK Query Hints in Plan Guides.

 Examples

A. Using MERGE JOIN

The following example specifies that the JOIN operation in the query is performed by MERGE JOIN.

Copy Code

USE AdventureWorks;

GO

SELECT * 

FROM Sales.Customer AS c

INNER JOIN Sales.CustomerAddress AS ca ON c.CustomerID = ca.CustomerID

WHERE TerritoryID = 5

OPTION (MERGE JOIN);

GO

B. Using OPTIMIZE FOR

The following example instructs the query optimizer to use the value 'Seattle' for local variable @city_name and to use statistical data to determine the value for the local variable @postal_code when optimizing the query.

Copy Code

USE AdventureWorks;

GO

DECLARE @city_name nvarchar(30);

DECLARE @postal_code nvarchar(15);

SET @city_name = 'Ascheim';

SET @postal_code = 86171;

SELECT * FROM Person.Address

WHERE City = @city_name AND PostalCode = @postal_code

OPTION ( OPTIMIZE FOR (@city_name = 'Seattle', @postal_code UNKNOWN) );

GO

C. Using MAXRECURSION

MAXRECURSION can be used to prevent a poorly formed recursive common table expression from entering into an infinite loop. The following example intentionally creates an infinite loop and uses the MAXRECURSION hint to limit the number of recursion levels to two.

Copy Code

USE AdventureWorks;

GO

--Creates an infinite loop

WITH cte (EmployeeID, ManagerID, Title) AS

(

    SELECT EmployeeID, ManagerID, Title

    FROM HumanResources.Employee

    WHERE ManagerID IS NOT NULL

  UNION ALL

    SELECT cte.EmployeeID, cte.ManagerID, cte.Title

    FROM cte 

    JOIN  HumanResources.Employee AS e 

        ON cte.ManagerID = e.EmployeeID

)

--Uses MAXRECURSION to limit the recursive levels to 2

SELECT EmployeeID, ManagerID, Title

FROM cte

OPTION (MAXRECURSION 2);

GO

After the coding error is corrected, MAXRECURSION is no longer required.

D. Using MERGE UNION

The following example uses the MERGE UNION query hint.

Copy Code

USE AdventureWorks;

GO

SELECT *

FROM HumanResources.Employee AS e1

UNION

SELECT *

FROM HumanResources.Employee AS e2

OPTION (MERGE UNION);

GO

E. Using HASH GROUP and FAST

The following example uses the HASH GROUP and FAST query hints.

Copy Code

USE AdventureWorks;

GO

SELECT ProductID, OrderQty, SUM(LineTotal) AS Total

FROM Sales.SalesOrderDetail

WHERE UnitPrice < $5.00

GROUP BY ProductID, OrderQty

ORDER BY ProductID, OrderQty

OPTION (HASH GROUP, FAST 10);

GO

F. Using MAXDOP

The following example uses the MAXDOP query hint.

Copy Code

USE AdventureWorks ;

GO

SELECT ProductID, OrderQty, SUM(LineTotal) AS Total

FROM Sales.SalesOrderDetail

WHERE UnitPrice < $5.00

GROUP BY ProductID, OrderQty

ORDER BY ProductID, OrderQty

OPTION (MAXDOP 2);

GO

G. Using INDEX

The following examples use the INDEX hint. The first example specifies a single index. The second example specifies multiple indexes for a single table reference. In both examples, because the INDEX hint is applied on a table that uses an alias, the TABLE HINT clause must also specify the same alias as the exposed object name.

Copy Code

USE AdventureWorks;

GO

EXEC sp_create_plan_guide 

    @name = N'Guide1', 

    @stmt = N'SELECT c.LastName, c.FirstName, e.Title

              FROM HumanResources.Employee AS e 

              JOIN Person.Contact AS c ON e.ContactID = c.ContactID

              WHERE e.ManagerID = 2;', 

    @type = N'SQL',

    @module_or_batch = NULL, 

    @params = NULL, 

    @hints = N'OPTION (TABLE HINT(e, INDEX (IX_Employee_ManagerID)))';

GO

EXEC sp_create_plan_guide 

    @name = N'Guide2', 

    @stmt = N'SELECT c.LastName, c.FirstName, e.Title

              FROM HumanResources.Employee AS e

              JOIN Person.Contact AS c ON e.ContactID = c.ContactID

              WHERE e.ManagerID = 2;', 

    @type = N'SQL',

    @module_or_batch = NULL, 

    @params = NULL, 

    @hints = N'OPTION (TABLE HINT(e, INDEX(PK_Employee_EmployeeID, IX_Employee_ManagerID)))';

GO

H. Using FORCESEEK

The following example uses the FORCESEEK table hint. Because the INDEX hint is applied on a table that uses a two-part name, the TABLE HINT clause must also specify the same two-part name as the exposed object name.

Copy Code

USE AdventureWorks;

GO

EXEC sp_create_plan_guide 

    @name = N'Guide3', 

    @stmt = N'SELECT c.LastName, c.FirstName, HumanResources.Employee.Title

              FROM HumanResources.Employee

              JOIN Person.Contact AS c ON HumanResources.Employee.ContactID = c.ContactID

              WHERE HumanResources.Employee.ManagerID = 3

              ORDER BY c.LastName, c.FirstName;', 

    @type = N'SQL',

    @module_or_batch = NULL, 

    @params = NULL, 

    @hints = N'OPTION (TABLE HINT( HumanResources.Employee, FORCESEEK))';

GO

I. Using multiple table hints

The following example applies the INDEX hint to one table and the FORCESEEK hint to another.

Copy Code

USE AdventureWorks;

GO

EXEC sp_create_plan_guide 

    @name = N'Guide4', 

    @stmt = N'SELECT e.ManagerID, c.LastName, c.FirstName, e.Title

              FROM HumanResources.Employee AS e 

              JOIN Person.Contact AS c ON e.ContactID = c.ContactID

              WHERE e.ManagerID = 3;', 

    @type = N'SQL',

    @module_or_batch = NULL, 

    @params = NULL, 

    @hints = N'OPTION (TABLE HINT ( e, INDEX( IX_Employee_ManagerID ) ) 

                       , TABLE HINT ( c, FORCESEEK) )';

GO

J. Using TABLE HINT to override an existing table hint

The following example shows how to use the TABLE HINT hint without specifying a hint to override the behavior of the INDEX table hint specified in the FROM clause of the query.

Copy Code

USE AdventureWorks;

GO

EXEC sp_create_plan_guide 

    @name = N'Guide5', 

    @stmt = N'SELECT e.ManagerID, c.LastName, c.FirstName, e.Title

              FROM HumanResources.Employee AS e WITH (INDEX (IX_Employee_ManagerID))

              JOIN Person.Contact AS c ON e.ContactID = c.ContactID

              WHERE e.ManagerID = 3;', 

    @type = N'SQL',

    @module_or_batch = NULL, 

    @params = NULL, 

    @hints = N'OPTION (TABLE HINT(e))';

GO

K. Specifying semantics-affecting table hints

The following example contains two table hints in the query: NOLOCK, which is semantic-affecting, and INDEX, which is non-semantic-affecting. To preserve the semantics of the query, the NOLOCK hint is specified in the OPTIONS clause of the plan guide. In addition to the NOLOCK hint, the INDEX and FORCESEEK hints are specified and replace the non-semantic-affecting INDEX hint in the query when the statement is compiled and optimized.

Copy Code

USE AdventureWorks;

GO

EXEC sp_create_plan_guide 

    @name = N'Guide6', 

    @stmt = N'SELECT c.LastName, c.FirstName, e.Title

              FROM HumanResources.Employee AS e 

                   WITH (NOLOCK, INDEX (PK_Employee_EmployeeID))

              JOIN Person.Contact AS c ON e.ContactID = c.ContactID

              WHERE e.ManagerID = 3;',

    @type = N'SQL',

    @module_or_batch = NULL, 

    @params = NULL, 

    @hints = N'OPTION (TABLE HINT ( e, INDEX( IX_Employee_ManagerID) , NOLOCK, FORCESEEK ))';

GO

The following example shows an alternative method to preserving the semantics of the query and allowing the optimizer to choose an index other than the index specified in the table hint. This is done by specifying the NOLOCK hint in the OPTIONS clause (because it is semantic-affecting) and specifying the TABLE HINT keyword with only a table reference and no INDEX hint.

Copy Code

USE AdventureWorks;

GO

EXEC sp_create_plan_guide 

    @name = N'Guide7', 

    @stmt = N'SELECT c.LastName, c.FirstName, e.Title

              FROM HumanResources.Employee AS e 

                   WITH (NOLOCK, INDEX (PK_Employee_EmployeeID))

              JOIN Person.Contact AS c ON e.ContactID = c.ContactID

              WHERE e.ManagerID = 2;',

    @type = N'SQL',

    @module_or_batch = NULL, 

    @params = NULL, 

    @hints = N'OPTION (TABLE HINT ( e, NOLOCK))';

GO