SQL Server Code,Tips and Tricks, Performance Tuning

Tuesday, May 02, 2006

A Developer’s Guide to SQL Server 2005

The follow up to the highly succesful book 'A First Look at SQL Server 2005 for Developers' has been published by Addison-Wesley
The book is 1008 pages long

From the site:

Few technologies have been as eagerly anticipated as Microsoft SQL Server 2005. Now, two SQL Server insiders deliver the definitive hands-on guide--accurate, comprehensive, and packed with examples. A Developer's Guide to SQL Server 2005 starts where Microsoft's documentation, white papers, and Web articles leave off, showing developers how to take full advantage of SQL Server 2005's key innovations. It draws on exceptional cooperation from Microsoft's SQL Server developers and the authors' extensive access to SQL Server 2005 since its earliest alpha releases.

You'll find practical explanations of the new SQL Server 2005 data model, built-in .NET hosting, improved programmability, SQL:1999 compliance, and much more. Virtually every key concept is illuminated via sample code that has been fully updated for and tested with the shipping version of the product.

Key coverage includes
Using SQL Server 2005 as a .NET runtime host: extending the server while enhancing security, reliability, and performance
Writing procedures, functions, triggers, and types in .NET languages
Exploiting enhancements to T-SQL for robust error-handling, efficient queries, and improved syntax
Effectively using the XML data type and XML queries
Implementing native SQL Server 2005 Web Services
Writing efficient, robust clients for SQL Server 2005 using ADO.NET, classic ADO, and other APIs
Taking full advantage of user-defined types (UDTs), query notifications, promotable transactions, and multiple active result sets (MARS)
Using SQL Management Objects (SMO), SQL Service Broker, and SQL Server Notification Services to build integrated applications

Download the Sample Chapter related to this title.

The Amazon link is here

Monday, May 01, 2006

Check For Not In Table Values (16 Different Ways And Counting)

How many times did you see a question like this:
how do I select all rows from a table where the column value is not 'ValueA' and not 'ValueB'
Well I decided to figure out how many different ways there are to do such a query
I came up with 16 different ways so far, some of them will cause a table scan, some of them will cause a index seek instead
You can test them out for yourself to see which ones perform the best

CREATE TABLE #TestTable (
Objects VARCHAR(20) PRIMARY KEY)

INSERT INTO #TestTable
VALUES ('boat')

INSERT INTO #TestTable
VALUES('ship')

INSERT INTO #TestTable
VALUES('car')

INSERT INTO #TestTable
VALUES('bus')

INSERT INTO #TestTable
VALUES ('airplane')

-- #1 <>
SELECT *,1
FROM #TestTable
WHERE Objects <> 'boat'
AND Objects <> 'ship'

-- #2 !=
SELECT *,2
FROM #TestTable
WHERE Objects != 'boat'
AND Objects != 'ship'

-- #3 NOT
SELECT *,3
FROM #TestTable
WHERE NOT (Objects = 'boat'
OR Objects = 'ship')

-- #4 NOT IN
SELECT *,4
FROM #TestTable
WHERE Objects NOT IN ('boat',
'ship')

-- #5 ALL
SELECT *,5
FROM #TestTable
WHERE Objects <> ALL (SELECT 'boat'
UNION ALL
SELECT 'ship')

-- #6 ANY
SELECT *,6
FROM #TestTable
WHERE NOT Objects = ANY (SELECT 'boat'
UNION ALL
SELECT 'ship')

-- #7 SOME
SELECT *,7
FROM #TestTable
WHERE NOT Objects = SOME (SELECT 'boat'
UNION ALL
SELECT 'ship')

-- #8 NOT IN with subquery
SELECT *,8
FROM #TestTable
WHERE Objects NOT IN (SELECT 'boat'
UNION ALL
SELECT 'ship')

-- #9 NOT EXISTS
SELECT *,9
FROM #TestTable T
WHERE NOT EXISTS (SELECT *
FROM (SELECT 'boat' AS Objects
UNION ALL
SELECT 'ship') F
WHERE F.Objects = T.Objects)

-- #10 LEFT OUTER JOIN
SELECT T.*,10
FROM #TestTable T
LEFT OUTER JOIN (SELECT 'boat' AS Objects
UNION ALL
SELECT 'ship') F
ON T.Objects = F.Objects
WHERE F.Objects IS NULL

-- #11 CHARINDEX
SELECT *,11
FROM #TestTable
WHERE CHARINDEX('-' + Objects + '-','-boat-ship-') = 0

-- #12 RIGHT OUTER JOIN
SELECT T.*,12
FROM (SELECT 'boat' AS Objects
UNION ALL
SELECT 'ship') F
RIGHT OUTER JOIN #TestTable T
ON T.Objects = F.Objects
WHERE F.Objects IS NULL
AND T.Objects IS NOT NULL

-- #13 FULL OUTER JOIN
SELECT T.*,13
FROM (SELECT 'boat' AS Objects
UNION ALL
SELECT 'ship') F
FULL OUTER JOIN #TestTable T
ON T.Objects = F.Objects
WHERE F.Objects IS NULL

-- #14 PATINDEX
SELECT *,14
FROM #TestTable
WHERE PATINDEX('%-%' + Objects + '%-%','-boat-ship-') = 0

-- #15 PARSENAME
SELECT *,15
FROM #TestTable
WHERE Objects <> PARSENAME('boat.ship',1)
AND Objects <> PARSENAME('boat.ship',2)

-- #16 REVERSE and NOT IN
SELECT *,16
FROM #TestTable
WHERE REVERSE(Objects) NOT IN ('taob','pihs')

-- And here is another one that BugsBunny suggested in a comment
-- #17 NOT LIKE
SELECT *,17
FROM #TestTable
WHERE ',' + 'boat,ship' + ',' NOT LIKE '%,' + Objects + ',%'

DROP TABLE #TestTable

So that's it, if you know of another way, leave me a comment or email me and I will add it to this list

Friday, April 28, 2006

Splitting City, State and Zipcode From One Column (Or Flat file)

Sometimes you deal with vendors, customers or agencies and you have to do a file exchange
Of course the format of these files is always dictated by these people so what do you do when a CityStateZip column contains values like “Long Island City, NY 10013” or like “Princeton,NJ 08536 “? But in your Database it is normalized of course, and you have 3 columns. You will have to use a combination of LEFT, LTRIM, SUBSTRING, REPLACE and RIGHT. I am taking into account that there could be spaces in the column or even spaces in the name (New York City) So let’s get started and see what we get

CREATE TABLE #TestCityStateZip (csz CHAR(49))
INSERT INTO #TestCityStateZip VALUES ('city ,st 12223')
INSERT INTO #TestCityStateZip VALUES ('New York City,NY 10028')
INSERT INTO #TestCityStateZip VALUES ('Princeton , NJ 08536')
INSERT INTO #TestCityStateZip VALUES ('Princeton,NJ 08536 ')
INSERT INTO #TestCityStateZip VALUES ('Long Island City, NY 10013')
INSERT INTO #TestCityStateZip VALUES ('Long Island City, NY 10013 ')
INSERT INTO #TestCityStateZip VALUES ('Long Island City , NY 10013')
INSERT INTO #TestCityStateZip VALUES ('Long Island City ,NY 10013 ')

SELECT LEFT(csz,CHARINDEX(',',csz)-1)AS City,
LEFT(LTRIM(SUBSTRING(csz,(CHARINDEX(',',csz)+1),4)),2) AS State,
RIGHT(RTRIM(csz),CHARINDEX(' ',REVERSE(RTRIM(csz)))-1) AS Zip
FROM #TestCityStateZip

GO

Tuesday, April 25, 2006

SQL Server 2005 SP1 Books Online (April 2006) For Download

SQL Server 2005 SP1 Books Online (April 2006) is available for download
Download an updated version of Books Online for Microsoft SQL Server 2005. Books Online is the primary documentation for SQL Server 2005. The April 2006 update to Books Online contains new material and fixes to documentation problems reported by customers after SQL Server 2005 was released. Refer to "New and Updated Books Online Topics" for a list of topics that are new or updated in this version. Topics with significant updates have a Change History table at the bottom of the topic that summarizes the changes. Beginning with the April 2006 update, SQL Server 2005 Books Online reflects product upgrades included in SQL Server 2005 Service Pack 1 (SP1).

The download is 121 MB

http://www.microsoft.com/downloads/details.aspx?FamilyID=be6a2c5d-00df-4220-b133-29c1e0b6585f&DisplayLang=en

Monday, April 24, 2006

How To Script Multiple Jobs Or Stored Procedures In SQL Server 2005 Management Studio

One of the biggest complaints of people who moved from SQL server 2000 to SQL Server 2005 is the inability to script multiple objects. Well that feature is still available only it’s kind of hidden
Click the Jobs or Stored Procedures folder in SQL Server Management Studio, and then hit the F7 key; this will bring up the Summary pane. Highlight all the Jobs or Stored Procedures that you want to script using a combination of Shift and Ctrl keys, then right click, Script Job/Stored Procedure as..., and then choose where to save this script to. An image of how to script Stored Procedures is below

Thursday, April 20, 2006

SQL Server 2005 Service Pack 1 Download Link

Yesterday I reported that Service Pack 1 was available for download but I never posted the link so here it is with a 24 hour delay

http://www.microsoft.com/downloads/details.aspx?displaylang=en&FamilyID=cb6c71ea-d649-47ff-9176-e7cac58fd4bc

Wednesday, April 19, 2006

Microsoft Releases SQL Server 2005 Service Pack 1

Microsoft Corp. today announced the availability of Microsoft® SQL Server™ 2005 Service Pack 1 (SP1), the product’s first major milestone since the launch of SQL Server 2005 only a few months ago. The release follows the remarkable reception, momentum and feedback offered by customers and partners in 92 countries. This has informed and accelerated Microsoft’s multiyear vision of Your Data, Any Place, Any Time. Microsoft SQL Server 2005 SP1 includes production-ready Database Mirroring functionality and the new SQL Server Management Studio Express, along with additional updates to SQL Server 2005 Express Edition targeted at users such as independent software vendors (ISVs) wanting to take advantage of greater functionality.

SP1 delivers production-ready Database Mirroring functionality for continuous availability. This complements the existing Always On Technologies in SQL Server 2005 such as failover clustering, database snapshots, snapshot isolation and log shipping. Since launch, Database Mirroring has been extensively tested by Microsoft and its thriving customer community to help ensure that it provides the high availability necessary for the largest customer deployments. Today more than 20 SQL Server customers have deployed Database Mirroring into production

SP1 extends business insight to smaller-scale customers and ISVs using the free SQL Server 2005 Express Edition, which includes SQL Server Reporting Services (SSRS) functionality, Full Text Search and the newly released SQL Server Management Studio Express. This optional set of capabilities is already receiving positive feedback from Microsoft’s ISV partners

SP1 also advances dynamic applications with the new SQL Server Management Studio Express, a simplified graphical management environment for SQL Server Express Edition. This tool builds on Microsoft’s commitment to extending the breadth of the SQL Server family to support all the data storage needs of Microsoft’s customers. It also complements the recently announced SQL Server Everywhere Edition, a lightweight and rich subset of capabilities found in other SQL Server editions, targeted for application embedded storage on clients. SQL Server Everywhere Edition is targeted to be available as a Community Technology Preview (CTP) this summer and released by year end.

For end-to-end business insight, SP1 extends SSRS to support enterprise reporting on SAP NetWeaver Business Intelligence with two new components in SP1: a Microsoft .NET Data provider for SAP NetWeaver Business Intelligence and a new MDX Query Designer. This new functionality extends the benefits of enterprise reporting with SQL Server by enabling SAP customers to easily create and manage reports on information inside any SAP BW data warehouse

Inside Microsoft SQL Server 2005: T-SQL Querying

How did I miss this? SQL Server MVP Itzik Ben-Gan has published his latest book: Inside Microsoft SQL Server 2005: T-SQL Querying. For all of you who read SQL Server magazine you probably know Itzik from his great SQL tips and puzzles articles.
Take a detailed look at the internal architecture of T-SQL—and unveil the power of set-based querying—with comprehensive reference and advice from the experts. Database developers and administrators get best practices, sample databases, and code to master the intricacies of the programming language—solving complex problems with real-world solutions.

Discover how to:

•Understand logical and physical query processing
•Apply a methodology to optimize query tuning
•Solve relational division problems
•Use CTEs and ranking functions to simplify and optimize solutions
•Aggregate data with various techniques, including tiebreakers, pivoting, histograms, and grouping factors
•Use the TOP option in a query to modify data
•Query specialized data structures with recursive logic, materialized path, or nested sets solutions
•PLUS—Improve your logic and get to the heart of querying problems with logic puzzles

The book is 640 printed pages, below is the table of contents

Chapter 01 - Logical Query Processing
Chapter 02 - Physical Query Processing
Chapter 03 - Query Tuning
Chapter 04 - Subqueries, Table Expressions and Ranking Functions
Chapter 05 - Joins and Set Operations
Chapter 06 - Aggregating and Pivoting Data
Chapter 07 - TOP and APPLY
Chapter 08 - Data Modification
Chapter 09 - Graphs, Trees, Hierarchies and Recursive Queries
Appendix A - Logic Puzzles

I could not find a sample chapter yet but as soon one is available I will create a link to it

If you are intereseted in purchasing this book the Amazon link is here

Tuesday, April 18, 2006

Use PARSENAME, CHARINDEX, PATINDEX or SUBSTRING To Grab Values Up To A Certain Character

This is a question that came up yesterday in the Getting started with SQL Server MSDN forum (http://forums.microsoft.com/MSDN/ShowPost.aspx?PostID=353250&SiteID=1)
A person wanted to use a MID function in SQL Server, There is no MID function in SQL Server but there are at least 4 ways to implement what the person tried to accomplish
Basically if the data looked like this

aaa-bbbbb
ppppp-bbbbb
zzzz-xxxxx

Then we need to grab everything up until the minus sign
So let's get started

CREATE TABLE #Test (
SomeField VARCHAR(49))

INSERT INTO #Test
VALUES ('aaa-bbbbb')

INSERT INTO #Test
VALUES ('ppppp-bbbbb')

INSERT INTO #Test
VALUES ('zzzz-xxxxx')

--using PARSENAME
SELECT PARSENAME(REPLACE(SomeField,'-','.'),2)
FROM #Test

--using LEFT and CHARINDEX
SELECT LEFT(SomeField,CHARINDEX('-',SomeField) - 1)
FROM #Test

--using LEFT and PATINDEX
SELECT LEFT(SomeField,PATINDEX('%-%',SomeField) - 1)
FROM #Test

--using CASE SUBSTRING and LEFT
SELECT CASE SUBSTRING(SomeField,4,1)
WHEN '-' THEN LEFT(SomeField,3)
ELSE LEFT(SomeField,4)
END
FROM #Test

--clean up
DROP TABLE #Test

Monday, April 17, 2006

Grant Execute/SELECT Permissions For All User Defined Functions To A User

You want to add a new user with read and write access and also the ability to execute all user defined functions but you don't want to make the user a db_owner. The code below will do a GRANT EXECUTE/SELECT for all the user defined functions in the DB If the user defined function is a table-valued function then you need to grant select permissions otherwise you need to grant execute permissions

Right now this code prints the GRANT EXECUTE/SELECT statements, change the PRINT to EXEC if you want it to be done automatically

--Grab all the functions for the current DB
SELECT IDENTITY(INT,1,1) AS ID,
SPECIFIC_NAME,DATA_TYPE
INTO #FunctionList
FROM INFORMATION_SCHEMA.ROUTINES --Only Procs
WHERE OBJECTPROPERTY(OBJECT_ID(SPECIFIC_NAME),'IsMSShipped') =0
AND ROUTINE_TYPE='FUNCTION'
ORDER BY SPECIFIC_NAME

DECLARE
@Loopid INT,
@MaxId INT,
@UserName VARCHAR(50)

--This is the user that will get the execute/select permissions
SELECT @UserName = 'SomeUser'

--Grab start and end values for the loop
SELECT @Loopid = 1,
@MaxId = MAX(ID)
FROM #FunctionList

DECLARE
@SQL VARCHAR(500),
@ProcName VARCHAR(400) ,
@Permission VARCHAR(20),
@DataType VARCHAR(20)

--This is where the loop starts
WHILE @Loopid <= @MaxId BEGIN

--grab the function name and type
SELECT @ProcName = SPECIFIC_NAME, @DataType =DATA_TYPE
FROM #FunctionList
WHERE ID = @Loopid

--Find out if it's a table-valued function
IF @DataType ='TABLE'
SELECT @Permission ='SELECT'
ELSE
SELECT @Permission ='EXECUTE'

--construct the statement
SELECT @SQL = 'GRANT ' + @Permission +' ON [' + @ProcName + '] TO ' + @UserName
PRINT (@SQL) --change PRINT to EXECUTE if you want it to run automatically

--increment counter
SET @Loopid = @Loopid + 1
END

--clean up
DROP TABLE #FunctionList

Thursday, April 13, 2006

Grant Execute Permissions For All Stored Procedures To A User

You want to add a new user with read and write access and also the ability to execute all stored procedures but you don't want to make the user a db_owner.
The code below will do a GRANT EXECUTE for all the procedures in the DB
This line will skip those dt_ procedures that are in every database
WHERE OBJECTPROPERTY(OBJECT_ID(SPECIFIC_NAME),'IsMSShipped') =0

Right now this code prints the GRANT EXECUTE statements, change the PRINT to EXEC if you want it to be done automatically

--Grab all the procedures for the current DB
SELECT IDENTITY(INT,1,1) AS ID,
SPECIFIC_NAME
INTO #Procedurelist
FROM INFORMATION_SCHEMA.ROUTINES --Only Procs
WHERE OBJECTPROPERTY(OBJECT_ID(SPECIFIC_NAME),'IsMSShipped') =0
AND ROUTINE_TYPE='PROCEDURE'
ORDER BY SPECIFIC_NAME

DECLARE
@Loopid INT,
@MaxId INT,
@UserName VARCHAR(50)

--This is the user that will get the execute permissions
SELECT @UserName = 'SomeUser'

--Grab start and end values for the loop
SELECT @Loopid = 1,
@MaxId = MAX(ID)
FROM #Procedurelist

DECLARE
@SQL VARCHAR(500),
@ProcName VARCHAR(400)

--This is where the loop starts
WHILE @Loopid <= @MaxId BEGIN

--grab the procedure name
SELECT @ProcName = SPECIFIC_NAME
FROM #Procedurelist
WHERE ID = @Loopid

--construct the statement
SELECT @SQL = 'GRANT EXECUTE ON ' + @ProcName + ' TO ' + @UserName
PRINT (@SQL) --change PRINT to EXECUTE if you want it to run automatically

--increment counter
SET @Loopid = @Loopid + 1
END

--clean up
DROP TABLE #Procedurelist

Wednesday, April 12, 2006

Use OBJECT_DEFINITION To Track Procedure Changes

Not everyone uses Visual Source Safe, CVS or Subversion to keep track of proc changes/deletions and/or additions
Of course you could use Red-Gate SQL Compare (I do) But let's say you don't have any of these tools and are using SQL Server 2005, what else can you do?
In SQL Server 2000 you can use
select ROUTINE_DEFINITION,SPECIFIC_NAME FROM INFORMATION_SCHEMA.ROUTINES to get the body of the stored procedure, the caveat is that this will only return 4000 characters.
Another way is to use the sp_helptext procedure
In SQL server 2005 this is much easier. There is a new function in town: OBJECT_DEFINITION()

OBJECT_DEFINITION() does return the whole body of a stored procedure
Below is some code (very simple) that will give you an idea of how you could use OBJECT_DEFINITION() to keep track of changes
You will have to setup a job that runs once a day and stores the definition of all the procedures in a table
Then you can do a self join on that table to find added, deleted and changed procedures
You can run the code below in 1 shot if you want

USE master
GO
--Let's Create a New Database
CREATE DATABASE TestProcCode
GO

USE TestProcCode
GO

--proc0
CREATE PROC proc0
AS
SELECT CURRENT_TIMESTAMP
GO

--proc1
CREATE PROC proc1
AS
SELECT GETDATE()
GO

--proc2
CREATE PROC proc2
AS
SELECT HOST_NAME()
GO

--create the proc changes table, bad name I know
CREATE TABLE ProcChanges (ID INT IDENTITY,RunDate DATETIME,ProcName VARCHAR(100),ProcCode VARCHAR(MAX))
GO

--insert all the procs that exist now
INSERT INTO ProcChanges
SELECT '20060410',SPECIFIC_NAME,OBJECT_DEFINITION( OBJECT_ID(SPECIFIC_NAME))
FROM INFORMATION_SCHEMA.ROUTINES
GO

--Let's change proc2
ALTER PROC proc2
AS
SELECT HOST_ID()
GO

--proc 3 is new
CREATE PROC proc3
AS
SELECT SUSER_SNAME()
GO

--proc 1 is deleted
DROP PROCEDURE proc1
GO

--insert all the procs that exist now
INSERT INTO ProcChanges
SELECT '20060411',SPECIFIC_NAME,OBJECT_DEFINITION( OBJECT_ID(SPECIFIC_NAME))
FROM INFORMATION_SCHEMA.ROUTINES
GO

--grab all deleted procs
SELECT 'Deleted',p1.*
FROM ProcChanges p1
LEFT OUTER JOIN ProcChanges p2 ON p1.ProcName =p2.ProcName
AND p2.RunDAte ='20060411'
WHERE p1.RunDAte ='20060410'
AND p2.ID IS NULL

--grab all added procs
SELECT 'Added',p1.*
FROM ProcChanges p1
LEFT OUTER JOIN ProcChanges p2 ON p1.ProcName =p2.ProcName
AND p2.RunDAte ='20060410'
WHERE p1.RunDAte ='20060411'
AND p2.ID IS NULL

--grab all changed procs
SELECT 'Changed',p1.*
FROM ProcChanges p1
JOIN ProcChanges p2 ON p1.ProcName =p2.ProcName
WHERE p1.RunDAte ='20060410'
AND p2.RunDAte ='20060411'
AND p1.ProcCode <> p2.ProcCode

--grab all procs that didn't change
SELECT 'Not Changed',p1.*
FROM ProcChanges p1
JOIN ProcChanges p2 ON p1.ProcName =p2.ProcName
WHERE p1.RunDAte ='20060410'
AND p2.RunDAte ='20060411'
AND p1.ProcCode = p2.ProcCode

USE MASTER
GO

--let's clean up this mess ;-)
DROP DATABASE TestProcCode
GO

What I have shown is very simple, you can expand on this and check for date ranges and improve on this a lot if you need to

Monday, April 10, 2006

NULL Is Not 'NULL'

Today someone posted a question on the Tek-Tips Forums web site
The queston was how not to insert rows with a NULL value in a certain column
The answer is of course INSERT INTO Table2 SELECT * FROM table WHERE column IS NOT NULL
This person replied that rows where the column is null are still being inserted
I turned out that there was data that had NULL character values 'NULL'
When you run SELECT NULL,'NULL' in Query Analyzer or SQL Server Managment Studio this looks identical so it's very easy to think that there is something else going on

Run the code below in Query Analyzer to understand what I mean

CREATE TABLE #test (
SomeField VARCHAR(50))

INSERT INTO #test
VALUES (NULL)

INSERT INTO #test
VALUES ('NULL')

SELECT *
FROM #test

SELECT *
FROM #test
WHERE SomeField IS NOT NULL

SELECT *
FROM #test
WHERE SomeField = 'NULL'

DROP TABLE #test

Friday, April 07, 2006

SQL Server Everywhere Edition <> Sybase SQL Anywhere

Paul Flessner Senior Vice President of Microsoft Corporation has posted a SQL Server 2005 update here
Am I the only one who thinks that the name "SQL Server Everywhere Edition" is very close to "Sybase SQL Anywhere"?
I can see the jokes already "Not only does SQL server have the same codebase (not true) the name is also ripped off"

Anyway these are the 4 key themes that Paul Flessner mentioned
Continuous Availability and Automation
Beyond Relational
Dynamic Applications
End-To-End Insight

This is what Paul has to say about SQL Server Everywhere Edition "This new offering for storage on clients of all types will provide a lightweight, compact, but rich subset of the capabilities found in other SQL Server editions. Beyond having rich local data management capabilities, SQL Server Everywhere Edition will also include support for seamlessly synchronizing with other SQL Server editions and provides features that promote building rich client applications that operate effectively in today’s increasingly “occasionally connected” environment. SQL Server Everywhere Edition also shares a common programming model with the other SQL Server editions, enabling developers to transfer skills and knowledge quickly and easily. We expect to ship the first CTP of SQL Server Everywhere Edition this summer, with the goal of final release before the end of this calendar year."

Wednesday, April 05, 2006

Left Join

As promised here is a left join blog post. What is a left (outer) join? This is what Books On Line has as the left join description “The Left Outer Join logical operator returns each row that satisfies the join of the first (top) input with the second (bottom) input. It also returns any rows from the first input that had no matching rows in the second input. The nonmatching rows in the second input are returned as null values. If no join predicate exists in the Argument column, each row is a matching row.”
So let’s see how that works

CREATE TABLE #Customer
(CustomerID INT ,
LastName VARCHAR(100),
FirstName VARCHAR(100))

CREATE TABLE #Order
(OrderID INT identity ,
Amount DECIMAL(12,3),
CustomerID INT)

INSERT INTO #Customer VALUES (1,'Gates','Bill')
INSERT INTO #Customer VALUES (2,'Woz','Steve')
INSERT INTO #Customer VALUES (3,'Ellison','Larry')

INSERT INTO #Order VALUES (12.99,1)
INSERT INTO #Order VALUES (13.45,1)
INSERT INTO #Order VALUES (56.45,3)

--Regular Join, CustomerID 2 is not returned
SELECT c.*,o.*
FROM #Customer c
JOIN #Order o ON c.CustomerID =o.CustomerID

--Left Join, CustomerID 2 is returned
SELECT c.*,o.*
FROM #Customer c
LEFT JOIN #Order o ON c.CustomerID =o.CustomerID

--Right Join, CustomerID 2 is returned
SELECT c.*,o.*
FROM #Order o
RIGHT JOIN #Customer c ON c.CustomerID =o.CustomerID

--Return all the people without an order
SELECT c.*,o.*
FROM #Customer c
LEFT JOIN #Order o ON c.CustomerID =o.CustomerID
WHERE o.CustomerID IS NULL

--Left Join, CustomerID 2 is not returned because the condition is in the WHERE clause
--The WHERE clause is applied to the #Order table instead of the #Customer table
SELECT c.*,o.*
FROM #Customer c
LEFT JOIN #Order o ON c.CustomerID =o.CustomerID
WHERE o.CustomerID < 3

--Left Join, CustomerID 2 is returned (as well as 3) because the condition is in the AND clause
SELECT c.*,o.*
FROM #Customer c
LEFT JOIN #Order o ON c.CustomerID =o.CustomerID
AND o.CustomerID < color="#009900">

--The 'correct' way, applying the WHERE condition to the #Customer table
SELECT c.*,o.*
FROM #Customer c
LEFT JOIN #Order o ON c.CustomerID =o.CustomerID
WHERE c.CustomerID < 3

DROP TABLE #Order,#Customer