by Xander Zelders, http://www.dotnet4all.

com
SQL Server developers factsheet

C
Data types String Functions (T-SQL) Date and Time functions (T-SQL)
type Size Range (from/to)  ASCII (character) : Returns the ASCII code value of the leftmost character of character  DATEADD (datepart , number, date): Returns a new datetime value based on adding an
 CHAR (int) : Converts the integer ASCII code int to a character interval to the specified date.
Exact numerics  CHARINDEX (search, expression, [start]) : Returns starting position (int) of first occurrence of the string search  DATEDIFF (datepart , number, date): Returns the number of date and time boundaries
within table or string expression starting from position start crossed between two specified dates.

N
bigint 8 bytes -9,223,372,036,854,775,808  DATENAME (datepart , date): Returns a character string representing the specified datepart
9,223,372,036,854,775,807  DIFFERENCE (expression1, expression2) : Returns the difference between the SOUNDEX values of the two
character expressions as an integer of the specified date.
bit 1 bit 0 to 1
 LEFT (expression, int): Returns part of character string expression starting at int characters from the left.  DATEPART (datepart , date): Returns an integer that represents the specified datepart of the
decimal -10^38 +1 to 10^38 –1  LEN (expression) : Returns the number of characters of the string expression, excluding trailing blanks. specified date.
 LOWER (expression) : Returns character expression after converting uppercase string to lowercase  DAY (date): Returns an integer representing the day datepart of the specified date.
int 4 bytes -2,147,483,648 to
 LTRIM (expression) : Returns a character string after removing all leading blanks.  GETDATE : Returns the current system date and time.
2,147,483,647
 NCHAR (int) : Returns the Unicode character with the given integer code.  MONTH (date): Returns an integer that represents the month part of a specified date.
money 8 bytes -922,337,203,685,477.5808  YEAR (date): Returns an integer that represents the year part of a specified date.
+922,337,203,685,477.5807  PATINDEX ('%pattern%', expression) : Returns starting position of the first occurrence of a pattern in a specified
numeric 19 bytes -10^38 +1 to 10^38 –1 expression, or zeros if the pattern is not found, on all valid text and character data types.

CO
 REPLACE (expression1,expression2,expression3) : Replaces all occurrences of the second given string
smallint 2 bytes -32,768 to 32,767 expression in the first string expression with a third expression. Dateparts Cursor Functions (T-SQL)
 QUOTENAME (character_string[, quote_character]) : Returns a Unicode string with the delimiters added to make
smallmoney 4 bytes -214,748.3648 to
+214,748.3647
the input string a valid Microsoft® SQL Server™ delimited identifier. Datepart Abbreviations  @@CURSOR_ROWS
tinyint 1 byte 0 to 255  REPLICATE (expression, int) : Repeats a character expression a specified number of times Returns the number of qualifying rows currently
 REVERSE (expression) : Returns the reverse of a character expression. year yy, yyyy in the last cursor opened on the connection.
Approximate numerics  RIGHT (expression, int): Returns part of character string expression starting at int characters from the right. quarter qq, q
month mm, m  @@FETCH_STATUS
 RTRIM (expression) : Returns a character string after removing all trailing blanks. Returns the status of the last cursor FETCH
float 8 bytes -1.79E + 308 to dayofyear dy, y
 SOUNDEX (expression) : Returns a four-character (SOUNDEX) code. day dd, d statement issued against any cursor currently
1.79E + 308  SPACE (int) : Returns a string of int spaces.
real 4 bytes -3.40E + 38 to week wk, ww opened by the connection.
3.40E + 38
 STR (float_expression[, length[, decimal]]) : Returns character data converted from numeric data. weekday dd  CURSOR_STATUS
Dates  STUFF (expression1, start, length, expression2 ) : Deletes a specified length (length) of characters from hour hh A scalar function that allows the caller of a
expression1 and inserts another set (expression2) at a specified starting point (start) of expression1. minute mi, n
second ss, s stored procedure to determine whether or not
datetime 8 bytes Jan 1, 1753 to Dec 31, 9999  SUBSTRING (expression, start, length) : Returns part of character, binary, text expression or image expression the procedure has returned a cursor and result
millisecond ms
starting from position start with length length set for a given parameter.
smalldatetime 4 bytes Jan 1, 1900 to Jun 6, 2079  UNICODE (char) : Returns the Unicode int value for the first character of char.
 UPPER (expression) : Returns a character expression after converting lowercase string to uppercase.

Type / performance Characteristics

Mathematical Functions (T-SQL)
Character Strings System Functions (T-SQL)
 ABS (Expression): Returns the absolute (positive) value of a numeric expression.
char Fixed-length non-Unicode character.  ACOS (Expression): Returns the angle, in radians, whose cosine is the specified float
Max 8000 characters  @@ERROR : Returns the error number for the last Transact-SQL statement executed. expression; also called arccosine.
varchar Variable-length non-Unicode data.  @@IDENTITY : returns the last-inserted identity value.  ASIN (Expression): Returns the angle, in radians, whose sine is the specified float
Max 8000 characters  @@ROWCOUNT : Returns the number of rows affected by the last statement.
varchar(max) Variable-length non-Unicode data expression. This is also called arcsine.
Max 2^31 characters (SQL 2005)
 @@TRANCOUNT : Returns the number of active transactions for the current connection.  ATAN (Expression): Returns the angle in radians whose tangent is a specified float
text Variable-length non-Unicode data.
 APP_NAME : Returns the application name for the current session if set by the application. expression. This is also called arctangent.
Max 2,147,483,647 characters  CASE : Evaluates a list of conditions and returns one of multiple possible result expressions.  ATN2 (Expression): Returns the angle, in radians, between the positive x-axis and the ray
Unicode Character Strings  CAST (expression AS data_type) / CONVERT : Converts an expression of one data type to another. from the origin to the point (y, x), where x and y are the values of the specified float
 COALESCE (expression [ ,...n ]) : Returns the first nonnull expression among its arguments. expressions.
nchar Fixed-length Unicode data.  CURRENT_TIMESTAMP : Returns the current date and time. ANSI SQL equivalent to GETDATE.  CEILING(Expression): Returns the smallest integer greater than, or equal to, the specified
Max 4000 characters  CURRENT_USER : Returns the name of the current user. Equivalent to USER_NAME().
nvarchar Variable-length Unicode data. numeric expression.
 DATALENGTH (Expression) : Returns the number of bytes used to represent any expression.  COS (Expression): Returns the trigonometric cosine of the specified angle, in radians.
Max 4000 characters
nvarchar(max) Variable-length Unicode data
 FORMATMESSAGE (msg_number , [param_value [,...n]]) : Constructs a message from an existing message in  COT (Expression): Returns the trigonometric cotangent of the specified angle, in radians.
Max 2^30 characters (SQL 2005) sys.messages and returns the formatted message for further processing.  DEGREES (Expression): Returns the corresponding angle in degrees for an angle specified
ntext Variable-length Unicode data.  GETANSINULL : Returns the default nullability for the database for this session. in radians.
Max 1,073,741,823 characters  HOST_ID : Returns the workstation identification number.  EXP (Expression): Returns the exponential value of the specified float expression.
Binary Strings  HOST_NAME : Returns the workstation name.  FLOOR (Expression): Returns the largest integer less than or equal to the specified numeric
 IDENT_INCR : Returns the increment value (returned as numeric (@@MAXPRECISION,0)) specified during the expression.
binary Fixed-length binary data. creation of an identity column in a table or view that has an identity column.
Max 8000 bytes  LOG (Expression): Returns the natural logarithm of the specified float expression.
 IDENT_SEED : Returns the seed value (returned as numeric(@@MAXPRECISION,0)) that was specified when an  LOG10 (Expression): Returns the base-10 logarithm of the specified float expression.
varbinary Variable-length binary data.
Max 8000 bytes
identity column in a table or a view that has an identity column was created.  PI : Returns the constant value of PI.
varbinary(max) Variable-length binary data.  IDENTITY : to insert an identity column into a new table  POWER (Expression, y): Returns the value of the specified expression to the specified
Max 2^31 bytes (SQL 2005)  ISDATE (expression): Determines whether an input expression is a valid date. power.
image Variable-length binary data.  ISNULL (expression , replacement_value) : Replaces NULL with the specified value.  RADIANS (Expression): Returns radians of the numeric expression, in degrees.
Max 2,147,483,647 bytes.  ISNUMERIC (expression): Determines whether an expression is a valid numeric type.  RAND : Returns a random float value from 0 through 1.
Other types  NEWID : Creates a unique value of type uniqueidentifier.  ROUND (numeric_expression ,length [,function ]): Returns a numeric value, rounded to the
 NULLIF (expression , expression) : Returns a null value if the two specified expressions are equal. specified length or precision.
cursor A data type for variables or stored  PARSENAME ('object_name',object_piece) : Returns the specified part of an object name. Parts of an object that
procedure OUTPUT parameters that  SIGN (Expression): Returns the positive (+1), zero (0), or negative (-1) sign of the specified
contain a reference to a cursor. can be retrieved are the object name, owner name, database name, and server name. expression.
sql_variant A data type that stores values of  PERMISSIONS ([objectid [,'column']]): Returns a value containing a bitmap that indicates the statement, object, or  SIN (Expression): Returns the trigonometric sine of the specified angle, in radians, and in an
various SQL Server 2005-supported column permissions of the current user. approximate numeric, float, expression.
data types, except text, ntext, image,  SESSION_USER : returns the user name of the current context in the current database.  SQRT (Expression): Returns the square root of the specified float value.
timestamp, and sql_variant.  STATS_DATE : Returns the date that the statistics for the specified index were last updated.  SQUARE (Expression): Returns the square of the specified float value.
table Is a special data type that can be  SYSTEM_USER : Allows a system-supplied value for the current login to be inserted into a table when no default
used to store a result set for  TAN (Expression): Returns the tangent of the input expression.
value is specified.
processing at a later time.
timestamp Is a data type that exposes
 USER_NAME ([ID]): Returns a database user name from a specified identification number.
automatically generated, unique
binary numbers within a database.
 by Xander Zelders, http://www.dotnet4all.com

C
Execution Plan Icons (2005)
SELECT (T-SQL) ALTER TABLE (T-SQL) Checklist for fast queries
The full syntax of the SELECT statement is complex, but the Modifies a table definition by altering, adding or dropping columns and
main clauses can be summarized as: constraints, or by disabling or enabling constraints and triggers. The full syntax  Avoid non-sargable WHERE-clauses. If possible rewrite them to sargable Arithmetic Nonclustered
is complex, but the main clauses can be summarized as: ones Expression Index Scan
SELECT [DISTINCT] [{TOP int | TOP int PERCENT}]
columns ALTER TABLE table  In the WHERE-clause use the least likely true AND expression first Assert
Nonclustered

N
[INTO new_table] {[ALTER COLUMN column_name {new_data_type  Avoid using OR in the WHERE-clause if not all colums have an index Index Seek
FROM table_source [(precision[, scale] )][NULL | NOT NULL]  Avoid using UNION if UNION ALL also does the trick Bitmap
Nonclustered
 Avoid using UNION of two subsets from the same table. Instead use OR in Index Spool
[[INNER |{{ LEFT | RIGHT | FULL }[ OUTER ]}] JOIN | {ADD | DROP} ROWGUIDCOL}] | ADD
the WHERE-clause Bookmark Nonclustered
table_source2 ON table_source.primairy_key = {[ <column_definition> ]| column_name AS Lookup Index Update
table_source2.foreign_key][,...n] computed_column_expression}[,...n]  Avoid using SELECT * FROM when only a few columns are needed. Try
Clustered Online Index
[WHERE search_condition] | [WITH CHECK | WITH NOCHECK] ADD to specify each column Index Delete Insert
[GROUP BY group_by_expression] { <table_constraint> }[,...n] | DROP {[CONSTRAINT]  Avoid using COUNT(*) to check the existence of a record. Instead use
Clustered Parameter Table
[HAVING search_condition] constraint_name EXIST Index Insert Scan
[ORDER BY order_expression [ASC | DESC]] | COLUMN column}[,...n] | {CHECK | NOCHECK}  Always try to use a WHERE-clause in your query to narrow the results Clustered
 Try to use the best performing operator as possible Remote Delete

CO
CONSTRAINT Index Scan
 DISTINCT : Specifies that only unique rows can appear in {ALL | constraint_name[,...n]} | {ENABLE | DISABLE}  Avoid using NOT IN. Instead use EXIST, NOT EXIST, IN or LEFT OUTER Clustered
JOIN with a check for a NULL condition Remote Insert
the result set. Null values are considered equal for the TRIGGER {ALL | trigger_name[,...n]}} Index Seek
purposes of the DISTINCT keyword  Avoid using IN when EXISTS is also possible Clustered
Remote Query
 TOP n [PERCENT] : Specifies that the first n rows are to  Avoid using IN when BETWEEN is also possible Index Update
be output from the query result set. If PERCENT is also  In case using IN try to order the list of values so that the most frequently
specified, the first n percent are output. Creating / Altering other objects found values are placed first
Collapse Remote Scan

 INTO new_table: Creates a new table and inserts the  Avoid using SUBSTRING in the WHERE-clause. If possible use LIKE Compute
Remote Update
resulting rows from the query into it Stored procedures: instead Scalar
CREATE PROCEDURE <name> AS <sql_statement>  Sometimes consider rewriting a query using a OR to multiple queries Concatenatio
 GROUP BY : Specifies the groups into which output rows ALTER PROCEDURE <name> AS <sql_statement> n
RID Lookup
are to be placed and, if aggregate functions are included in combined with a UNION ALL
DROP PROCEDURE <name>  Don't use ORDER BY if you don't really need it Constant Row Count
the SELECT clause <select list>, calculates a summary Scan Spool
value for each group.  Keep the width and/or number of sorted columns to the minimum
Indexes:  Keep the number of rows to be sorted to a minimum Delete Segment
 HAVING : Specifies a search condition for a group or an CREATE INDEX <name> ON <table> (<column>)
aggregate  When sorting a specific column often conside making that column a
CREATE UNIQUE CLUSTERED INDEX <name> ON <table>.<column> Deleted Scan Sequence
clustered index
DROP INDEX <table>.<name>
 In case of using HAVING try to minimize the amount of rows using a Sequence
WHERE clause Eager Spool
Project
UPDATE (T-SQL) Views: 
CREATE VIEW <name> [(<Column1>,…)] AS <SELECT_statement> In case using LIKE on CHAR of VARCHAR colums quite often consider
Filter Sort
The full syntax of the UPDATE statement is complex, but the ALTER VIEW <name> [(<Column1>,…)] AS <SELECT_statement> using the full-text search option
main clauses can be summarized as: DROP VIEW <name>  In case using GROUP BY without an aggregate function try using
Hash Match Split
DISTINCT instead
UPDATE table_name  Avoid using variables in a WHERE clause in case the query is located in a Hash Match
SET column_name = {expression | DEFAULT | NULL} Triggers: Spool
batch-file Root
[,...n] CREATE TRIGGER <name> ON <table> FOR INSERT, UPDATE, Hash Match Stream
[WHERE <search_condition>] DELETE AS <sql_statement> Team Aggregate
ALTER TRIGGER <name> ON <table> FOR UPDATE, DELETE AS
<sql_statement> Checklist for creating indexes Insert Switch
DROP TRIGGER <name> Inserted
DELETE (T-SQL)  Create indexes on the highly selective colums that are used in the Scan
Table Delete
The full syntax of the DELETE statement is complex, but the Functions: WHERE-clause Iterator
main clauses can be summarized as: Table Insert
CREATE FUNCTION <name> RETURNS <data_type> AS  Create indexes on all columns that are used in the WHERE clause in case Catchall
RETURN <sql_expression> OR is used
DELETE [FROM] table_name Lazy Spool Table Scan
CREATE FUNCTION <name> RETURNS <data_type> AS BEGIN  Create at least a clustered index on every table. Generally use the column
[WHERE <search_condition>] <sql_statement> RETURN <sql_expression> END Log Row
that monotonically increases Scan
Table Spool
ALTER FUNCTION <name> RETURNS <data_type> AS  Create indexes columns that are frequently accessed by WHERE,
RETURN <sql_expression> Merge
ORDER BY, GROUP BY, TOP and DISTINCT Interval
Table Update
DROP FUNCTION <name> 
INSERT (T-SQL) Only add indexes that will be used frquently
Merge Join
Table-valued
INSERT adds a new row to an existing table or a view.  Avoid adding too much indexes on dynamic tables (subject to many Function
The full syntax of the INSERT statement is complex, but the INSERTs, UPDATEs or DELETEs) Nested
Sargability  For static tables use a FILLFACTOR and PAD_INDEX of 100. For Loops
Top
main clauses can be summarized as: SQL Server only uses indexes for colums used in sargable dynamic tables use a lower FILLFACTOR Nonclustered
expressions. Green = fastest expression.  To identify additional indexes use the SS Profiler Create Trace Wizard and UDX
INSERT [INTO] table_name Index Delete
[(column_list)] Sargable Non-sargable trace "Identify Scans of Large Tables" Nonclustered
VALUES ({ DEFAULT | NULL | expression }[,...n]) = IS NULL  Avoid adding indexes twice Index Insert
Update
> <>
Factsheet by Red: Temporary tables/spools
< != Xander Zelders
CREATE TABLE (T-SQL) Try rewriting the query
>= !> http://www.dotnet4all.com Red: Index or table scans
Creates a new table. The full syntax is complex, but the main <= !< Create additional or better indexes
clauses can be summarized as:
EXIST NOT Connectionstring Orange: Bookmark lookups
CREATE TABLE IS NOT EXIST  Consider changing the clustered index
[{database_name.[owner].| owner.}] table_name  SQL Server, Standard security:  Consider using a covering index
IN NOT IN Provider=sqloledb;Data Source=myServerAddress; Initial
({<column_definition> | column_name AS Catalog=myDataBase;User Id=myUsername;Password=myPassword;
 Consider limiting the number of
BETWEEN NOT LIKE columns in the SELECT statement.
computed_column_expression | <table_constraint>}  SQL Server, Trusted connection: Provider=sqloledb;Data Source=myServerAddress; Initial
[,...n]) LIKE 'abc%' LIKE '%abc'
Catalog=myDataBase;Integrated Security=SSPI;
Yellow: Filter and/or sort
LIKE '%abc%'  SQL Server 2005, Standard security: Driver={SQL Native Client};
 Consider removing any functions in the
function on column Server=myServerAddress;Database=myDataBase;Uid=myUsername;Pwd=myPassword;
WHERE clause
 Consider not using views in your code
column1 = column1  SQL Server 2005, Trusted connection: Driver={SQL Native Client};  Consider using additional indexes
column1 = column2 Server=myServerAddress;Database=myDataBase;Trusted_Connection=yes;  Consider not to sort