Tài liệu Bust a Move with Your SSIS – Passing Package Variables docx

Bust a Move with
Your SSIS –
Passing Package
Variables
1-800-COURSES
www.globalknowledge.com
Expert Reference Series of White Papers
Introduction
Integration Services (SSIS), the next generation of the Extract, Transport, and Load (ETL) feature included with
Microsoft SQL Server 2005 has befuddled Database Administrators. This is due to the new programming para-
digm and the complexity of the development environment. SSIS includes many new capabilities bundled with a
Visual Studio front end. This paper explores the creation of sample development data featuring the use of the
most basic features in this new interface. The challenge of declaring and passing package variables contains
enough material for an hour’
s skill-building session with this new tool.
Extract,Transport, and Load (ELT)
The challenge of the new changes of the ETL feature is comparable to a rock band taking over a square dance.
It’
s time to let go, learn to dance to a new beat, and bust some moves with this great new tool, SSIS. The com-
plexities of this new tool have caused many administrators to retreat to the familiar tempo of DTS. However,
as we explore this tool, you will find the creativity available in this new environment compelling enough to
bust some moves of your own.
The goal of this paper is to demonstrate the generation of a table of random data to be used in a development
environment, exposing the nuances of passing variable values into various tasks objects that make up the
package.
One of the challenges to application development in a database is to have data available for testing and
reporting. Adding to this challenge are the typical privacy requirements preventing use of actual patient per-
sonal identification. In my experience, it is always better to develop with test data that is as reflective of the
test data as possible, especially when interviewing end users using a prototype of the system under develop-
ment. I will walk thru the steps needed to create a SSIS package to develop a table of test data.
The end result of the execution of the package, the Person table, contains just a couple of challenges. The

package will be configured with package-level variables that will be passed from one task to another. This
poses the challenges of declaring the v
ariables and referencing them throughout the package
.
Our goal is to
generate random Patient Name and SSN information to be inserted in the Person Table schema shown below
in Step 1. This purpose of this table is to replace a table containing an extract of live production data to facili-
tate application testing. Overwriting the contents of a table with randomly generated data will obscure the pri-
vate details in a database, which are protected by legal constraints, while allowing proper acceptance testing
and comparison to an existing system prior to cut over to the new system.
Bill Kenworthy, Global Knowledge Instructor, MCDBA
Bust a Move with Your SSIS – Passing
Package Variables
Copyright ©2007 Global Knowledge T
raining LLC. All rights reserved.
Page 2
T
he project at hand starts with generation of a simple Database containing five tables and two stored proce-
dures. A SSIS solution is created to use this database and populate the Person table with random data in three
columns: FirstName, LastName, and SSN.
The name data is generated from the contents of two driving tables, FName and LName. These tables contain
the seed data for name generation. The result of the execution of the name generation is a table containing an
Identity column and columns for First and Last Name. The SSN generation fills a separate table and requires no
seed data.
The name generation takes place inside a For Next Loop that uses a package variable to control the number of
rows generated. Once the loop completes, execution is passed to a SQL task that runs the SSN generation
stored procedure. After completion of this task, a Merge Join Task is the last major data manipulation activity.
The Merge Join, contained in a Data Flow Task, combines the two staging tables into the final product. The
moves that make this package possible are creating package variables, passing the variable values to the task
objects, and careful attention to matching column data types. A data dictionary for the development database

is contained in Resource
A.
The Project
1. Generate tables and stored procedures for the project
2. Creating a SSIS project adding needed v
ariables
3.
Add a SQL Task truncating working tables
4. Add a For Next Loop configure looping parameters
5. Call a stored procedure in the loop passing a parameter.
6. Generating the SSN data using a execute sql task.
7. Add a data flow task
8. Configure a data merge task
The tasks to be accomplished in this project
The following the individual steps of creating the package;
howev-
er, the reader will find the configuration entries in the screen shots
are valuable in duplicating this demonstration. The database and
its objects are the foundation upon which we build our transforma-
tion.
The solution requires 2 seed tables and 2 staging tables used
to hold temporary results and a final table holding the merged con-
tents of the two staging tables. The database diagram in the data-
base is shown in Figure 1.
Step 1.
Generate tables and stored procedures for the
project
The data dictionary for this database is contained in the Resources
section of this document. The script for generating the schema and
stored procedures is in Resource B. The code has hard-coded refer-

ences to the DEV database;
the script should be run in the context
of a database with that name.
Copyright ©2007 Global Knowledge T
raining LLC. All rights reserved.
Page 3
Figur
e 1. Database Schema for the project
S
tep 2. Open an Integration
Services Project.
Open your project, then right click on
any clear space on the control flow pane
and choose variables from the context
menu to open the variable declaration
dialog box. Add two int32 variables,
Counter and MaxRows, with values of 0
and 1000, as shown in Figure 2. The
Counter variable is used to pass the cur-
rent loop index into the SQL task contained in the loop task that will be added to the project in step 4.
MaxRows is the number of rows to be inserted into the Person Table. Figure 3 shows the dialog box with
appropriate entries.
Note: references to variables in this
environment are case sensitive
.
Resource C contains a reference a topic
in the SQL Server 2005 Books Online
describing variables and links to how-to:
topics.
Step 3. Add a SQL Task truncating

working tables
Add a SQL
Task to the Control Flow win-
dow as the first task in the project and
set its parameters as shown in the dia-
gram. Note the configuration of the
ConnectionType as ADO.NET. Although
this SQL Task doesn’t pass parameters in
the SQLStatement property, I like to keep
settings of similar objects consistent.
Specifying ADO.NET as the connection
type allows reference to parameters
using the @ naming convention.
T
he SQL
query simply truncates the P
erson and Name tables. An appropriate reference to this object in the books online
is listed in Resource C.
Step 4. Add a For Next Loop configure looping parameters
T
he F
or Loop container defines a repeating control flow in a package
.
In this package
, the For loop is used to
repeat the execution of the Mak
eNames stored procedure until the required number of rows configured in the
MaxRow variable are inserted into the Name table. The For Loop container uses three elements to define the
loop init, eval, and assign(increment) control values. As you can see in Figure 4 above, the variable @Counter
is used for indexing in the loop. This reference is case-sensitive and must match a package variable name, with

the @ prefix necessary in this property page. For example, the variable @MaxRows matches the MaxRows
package variable
.
An appropriate reference to this object in the books online is listed in Resource C.
Copyright ©2007 Global Knowledge T
raining LLC. All rights reserved.
Page 4
F
igure 2. Variables dialog box
Figure 3. SQL Task to truncate the working tables
Figure 4. Configuring the For Loop Task
Step 5. Call a stored procedure in the loop passing a parameter.
SQL Task configured with connection type ADO.NET, calls stored procedure MakeAName. Note my preference
for the property, ConnectionT
ype
. Each connection type supports a different syntax for passing parameters.
ADO.NET supports the @ reference, other connection type use a ? [question mark]. I prefer the @ syntax, it is
consistent with the syntax used in Transact SQL. An appropriate reference to this object in the books online is
listed in Resource C.
Figure 5. Property page for the SQL Task embedded in the Loop Task
Copyright ©2007 Global Knowledge T
raining LLC. All rights reserved.
Page 5
Figure 6. Parameter map entry passes variable value from loop.
The second part of configuring this SQL Task is the mapping required to tie the variable referenced in the SQL
statement to the package variable value being passed into the SQL Task by its parent container. The first three
columns are selected from combo box choices, the developer enters the Parameter Name value by hand.
Figure 7. Control Flow diagram of the project to this point.
Test it!
Now the project is at a point where it can be tested. Your package should resemble the package shown in the

figure above. If your package errors out when you run the debugger, consult the Execution/Results view for
error messages and resolve the errors.
Copyright ©2007 Global Knowledge T
raining LLC. All rights reserved.
Page 6
S
tep 6 Generating the SSN data using an Execute SQL task.
Figure 8. Configuration of the SQL Task that follows execution of the loop
Step 7. Add a data flow task
Assemble the data flow objects as shown in Figure 8 above; the properties to be set are in the table in
Resource D. Appropriate references to the objects used in this dataflow for lookup in the books online are list-
ed in Resource C. The Merge Join Task properties are detailed in Step 8.
Figure 9. The Data Flow contains 6 objects
Copyright ©2007 Global Knowledge T
raining LLC. All rights reserved.
Page 7
S
tep 8. Configure a data merge task
This entire configuration of the Merge Join is shown in Figure 10. This is the only property page for the Merge Join.
Data typing is strong in this environment. The datatype of each column in the output table must match that of the
corresponding column in the P
erson table. The figure shows the FirstName output column has been configured as
a Unicode string, the LastName column in this datareader and the SSN column in the SSN datareader should be
set to Unicode as well.
Copyright ©2007 Global Knowledge T
raining LLC. All rights reserved.
Page 8
Figure 10. Configuration of the
Merge Join Object
Figure 11. Configuring the

datar
eader column datatype
properties
Figure 12. Final control flow of the project
The finished project should have a Control Flow diagram as shown in Figure 12. In this example annotations
have been added to label each task in the flow.
The diagram shows a few of the
rows from the Person table popu-
lated using the SSIS package. A
weakness in my calls to the
RAND() SQL function inside the
Mak
eSSN procedure shows a lot of
commonality in the second and
third segments of the data in the
SSN column. In this snapshot of
the data,
you see the value of 71 is
very popular in the second seg-
ment of the string, and a modal
distribution in the last four charac-
ters of the string. There are clumps
of similar v
alues
,
‘7137’ shows up
in rows 98 -103.
I think the inclu
-
sion of a Common Language

Runtime (CLR) Assembly with a
function to generate a random SSN
string would be a significant
improvement to the package and
provide a performance increase
.
Copyright ©2007 Global Knowledge T
raining LLC. All rights reserved.
Page 9
Figure 13. The data generated by
the package.
Summary
I’ve presented a common development scenario, generating representation test data and a possible solution to
this requirement. The solution presented demonstrates a control flow containing a looping task, several SQL
tasks, and a dataflow using a merge object. Declaring package variables and passing variable values between
tasks requires careful attention to detail when configuring the various tasks to share values amongst them.
SSIS presents a flexible programming structure allowing no practical limit to extension. This flexibility brings
with it a finer structure for controlling a group of tasks the complexity of which bears careful experimentation.
The environment provides many opportunities for tapping into the power of the .NET Framework but brings
with it some new baggage such as case sensitivity, connection type requirements, and strict type casting.
Learn More
Learn more about how you can improve productivity
, enhance efficiency, and sharpen your competitive edge.
Check out the following Global Knowledge courses:
Implementing and Maintaining Microsoft SQL Server 2005 Integration Services
Microsoft Certified IT Professional: Database Administrator Boot Camp
SQL Server 2005 Administration
SQL Server 2005 for Business Intelligence
SQL Server 2005 for Developers
SQL Server 2005 for Reporting Services

For more information or to register, visit www.globalknowledge.com or call 1-800-COURSES to speak with a
sales representative.
Our courses and enhanced, hands-on labs offer practical skills and tips that you can immediately put to use.
Our expert instructors draw upon their experiences to help you understand key concepts and how to apply
them to your specific work situation.
Choose from our more than 700 courses, delivered through Classrooms,
e-Learning,
and On-site sessions, to meet your IT and management training needs.
About the Author:
Bill Kenworthy has been working with SQL Server since version 6.0. His love for database challenges is reflect-
ed in his writing.
Bill lives with his wife and 2 dogs at the end of a dirt road in northern
Washington State.
Resources:
A. Data Dictionary for the project
Staging Tables
FName
,
LName
Two seed tables – number of rows not necessarily equal. These two tables contain the first
and last name values that will be selected randomly and inserted into a row in the Name
table.
Name ,SSN
Working tables holding Name and SSN working data
Copyright ©2007 Global Knowledge T
raining LLC. All rights reserved.
Page 10
P
roduction Table
Person

Stores Patient Name and SSN data
Stored procedures
MakeAName, requires an integer variable that is used to seed the RAND() function. The procedure
inserts a row into the Dev.dbo.Person table, providing values for the FirstName and LastName
columns. The name values are randomly selected from the staging tables.
MakeASSN, populates the SSN table with a unique combination of characters generated by the
RAND() function. The stored procedure checks the size of the Person table and inserts the
same number of rows into the staging table.
B. Script for creation of the database objects
SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
IF NOT EXISTS (SELECT * FROM sys.objects WHERE object_id =
OBJECT_ID(N'[dbo].[FName]') AND type in (N'U'))
BEGIN
CREATE TABLE [dbo].[FName](
[Id] [int] IDENTITY(1,1) NOT NULL,
[F
irstName] [nvarchar](50) NULL
) ON [PRIMARY]
END
GO
SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
IF NOT EXISTS (SELECT * FROM sys.objects WHERE object_id =
OBJECT_ID(N'[dbo].[LName]') AND type in (N'U'))
BEGIN

CREATE TABLE [dbo].[LName](
[Id] [int] IDENTITY(1,1) NOT NULL,
[LastName] [nvarchar](50) NULL
) ON [PRIMARY]
END
GO
SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
IF NOT EXISTS (SELECT * FROM sys.objects WHERE object_id =
OBJECT_ID(N'[dbo].[SSN]') AND type in (N'U'))
BEGIN
CREATE TABLE [dbo].[SSN](
[Id] [int] IDENTITY(1,1) NOT NULL,
[SSN] [char](11) NULL
) ON [PRIMARY]
Copyright ©2007 Global Knowledge T
raining LLC. All rights reserved.
Page 11
E
ND
GO
SET ANSI_NULLS ON
GO
S
ET QUOTED_IDENTIFIER ON
GO
IF NOT EXISTS (SELECT * FROM sys.objects WHERE object_id =
OBJECT_ID(N'[dbo].[Name]') AND type in (N'U'))

BEGIN
CREATE TABLE [dbo].[Name](
[PersonID] [int] IDENTITY(1,1) NOT NULL,
[FirstName] [varchar](50) NULL,
[LastName] [varchar](50) NULL
) ON [PRIMARY]
END
GO
SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
IF NOT EXISTS (SELECT * FROM sys.objects WHERE object_id =
OBJECT_ID(N'[dbo].[Person]') AND type in (N'U'))
BEGIN
CREATE TABLE [dbo].[Person](
[PersonID] [int] IDENTITY(1,1) NOT NULL,
[FirstName] [nvarchar](50) NULL,
[LastName] [nvarchar](50) NULL,
[SSN] [nchar](11) NULL
) ON [PRIMARY]
END
GO
SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
IF NOT EXISTS (SELECT * FROM sys.objects WHERE object_id =
OBJECT_ID(N'[dbo].[MakeAName]') AND type in (N'P', N'PC'))
BEGIN

EXEC dbo.sp_executesql @statement = N'
CREATE procedure [dbo].[MakeAName]
@counter int
as
Declare @fid int
Declare @lid int
declare @NumFirstNames int
declare @NumLastNames int
Declare @random int
Declare @fname varchar(50)
Declare @Lname varchar(50)
select @NumFirstNames = count(*) from fname
select @NumLastNames = count(*) from lname
set @counter = @counter *131
Copyright ©2007 Global Knowledge T
raining LLC. All rights reserved.
Page 12
i
f @counter is null
begin
set @counter = 99999
end
S
et @fid = @NumFirstNames - cast(rand(@counter) * @NumFirstNames
as int)
Set @lid = @NumLastNames - cast(rand(@counter) * @NumLastNames
as int)
Get a first name and last name pair using a random pointer
pair
select @fname = FirstName, @lname = LastName

from fname,lname
where fname.id = @fid and lname.id = @lid
Use the randomly paired first and lastname to insert a row
Insert Dev.dbo.Name (FirstName,LastName)
values (@fname,@lname)
'
END
GO
SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
IF NOT EXISTS (SELECT * FROM sys.objects WHERE object_id =
OBJECT_ID(N'[dbo].[MakeSSN]') AND type in (N'P', N'PC'))
BEGIN
EXEC dbo.sp_executesql @statement = N'
CREATE Procedure [dbo].[MakeSSN] as
declare @counter int
declare @X int
declare @Y int
declare @Z1 int
declare @Z2 int
declare @maxrows int
declare @a char(3)
declare @b char(2)
declare @c char(4)
declare @SSN char(11)
declare @test int
truncate table dbo.SSN
truncate table dbo.test

Select @maxrows = count(*) from dbo.Name
Set @counter = 0
set @x = 111
While @counter < @maxrows
Begin
Set @test = 1
if (@x < 842) set @x = @x + 3
else set @x = 111
Copyright ©2007 Global Knowledge T
raining LLC. All rights reserved.
Page 13
S
et @Y = cast(rand(@counter % 7 ) * 100 as int)
Set @Z1 = cast(rand(@counter % 30) * 10000 as int)
Set @a = cast(@x as varchar(3))
S
et @b = cast(@y as varchar(2))
Set @c = cast(@z1 as varchar(4))
set @ssn = @a + ''-'' + @b + ''-'' + @c
select @test = count(SSN) From SSN where SSN = @SSN
if @test = 0
Begin
if len(@ssn) = 11
begin
insert ssn (ssn) values (@ssn)
set @counter = @counter + 1
end
print ''too short''
end
print @counter

print @ssn
end
'
END
C.Topics in the SQL Server Books Online referenced in the text.
Step # Topic
1 Using Variables in Packages
2 Execute SQL Task
3 For Loop Container
4 Execute SQL Task
5 Execute SQL Task
6 Execute SQL
Task
7 DataReader Source
7
Sort
Tranformation
7 Merge Join Transformation
7 SQL Server Destination
Copyright ©2007 Global Knowledge T
raining LLC. All rights reserved.
Page 14
D.Table of Settings for the Dataflow Objects
Copyright ©2007 Global Knowledge T
raining LLC. All rights reserved.
Page 15
Object Tab Property Value
Name Datareader General ICbConnection Connection manager of
y
our creation

Component Properties Name Name Datareader
SQL Command Select * from dbo.DEV
Input and Output Properties IDataReaderOutput -
Output Columns -
FirstName
Unicode string [DT_WSTR]
DataReaderOutput -
Output Columns -
LastName
DataReaderOutput -
Output Columns -
LastName
Unicode string [DT_WSTR]
SSN Datareader General IDbConnection Connection manager of
your creation
Component Properties Name SSN Datareader
SQL Command Select * from dbo.SSN
Input and Output Properties DataReaderOutput -
Output Columns -
SSN
Unicode string [DT_WSTR]
Name Sort Input Column PersonID
Output Alias
Sort Type
Sort Order
SSN Sort Input Column
Output Alias
Sort
Type
Sort Order

Merge Join See the figure in Step 8
SQL Server Destination Connection Manager Connection of your creation
Mappings Mappings are straight
across, no map to the ˆD
column on the person table