Dynamic ETL with SSIS

Question

Dynamic ETL with SSIS

Sarah

Old Hand

Points: 352
More actions
October 18, 2010 at 10:20 pm

#229786

Comments posted to this topic are about the item Dynamic ETL with SSIS

Viewing 15 posts - 1 through 15 (of 33 total)

You must be logged in to reply to this topic. Login to reply

Martyn Hughes SSC Journeyman Points: 77 More actions · Answer 1

I don't know why you'd want to go to such lengths to use SSIS, just use Pervasive (formerly Data Junction). It's a far superior way of getting data into SQL.

Samuel Vella SSCrazy Eights Points: 8055 More actions · Answer 2

Martyn Hughes (10/19/2010)
I don't know why you'd want to go to such lengths to use SSIS, just use Pervasive (formerly Data Junction). It's a far superior way of getting data into SQL.

I agree with you to a point but there are more elegant ways of doing this in SSIS

Sarah, you mention that SSIS requires a flat file format to load the data properly in to SQL Server... so why not create the bulk load format file on the fly using a script? its fairly easy to parse the file and determine number of columns, data types etc. and then build the format file and the destination table based on that information.

Virendra Chauhan SSC Veteran Points: 274 More actions · Answer 3

Great work, if you have a spare time to do so. I think SSIS is designed to take care the work done in the back end. Storing the file name and others make sense but rest is really confusing for anyone who have to maintain or troubleshoot the package. I would say simple is better.

Nitya Say Hey Kid Points: 699 More actions · Answer 4

The assumption that files coming in are of fixed length is little hard to generalize.

In that case the Configuration table values are hard to determine.

Great work in getting it up and running

Steve Jones - SSC Editor SSC Guru Points: 734416 More actions · Answer 5

Martyn Hughes (10/19/2010)
I don't know why you'd want to go to such lengths to use SSIS, just use Pervasive (formerly Data Junction). It's a far superior way of getting data into SQL.

Not everyone can afford third party tools, or get the approval to buy them. Knowing how they could perform this in SSIS is a piece of knowledge that many can benefit from.

If there's a superior way to use Pervasive, perhaps you would want to mention specifically why rather than just add a snarky comment.

ranjith_s 65449 Mr or Mrs. 500 Points: 562 More actions · Answer 6

Thanks and nice work Sarah! But can you update it with link to source code. I know all the SQL scripts are in the content, but we need SSIS files too. Thanks in advance!

amenjonathan SSCrazy Points: 2482 More actions · Answer 7

Nitya (10/19/2010)
The assumption that files coming in are of fixed length is little hard to generalize.
In that case the Configuration table values are hard to determine.
Great work in getting it up and running

It's not a generalization if your suppliers have given you a data dictionary. Many of our data suppliers do the same. The files they give us are always formatted with these fixed field lengths in mind, which we must then TRIM before using properly. The only difference in our packages is we handle erroneous row lengths in the package.

Very interesting way of importing flat files. We also build specific packages for each of our files coming in from different suppliers, and it's a real pain. SSIS can handle a lot of different scenarios, but IMO it's not a simple enough process, as you have mentioned.

Good article. ... Need to think about the pros and cons of adopting it here. Thanks for sharing!

-------------------------------------------------------------------------------------------------
My SQL Server Blog

Martyn Hughes SSC Journeyman Points: 77 More actions · Answer 8

Steve, I've been a great fan of SSC for many years, I've used a number of redgate tools and Litespeed to compress some rather large DBs especially when using a SAN. I think Susan made a good contribution to the forum.

My comment is not meant to be 'snarky', I'm always troubled when I have to fight within an organization to get useful tools. Pervasive does cost several thousand but the amount of time saved using appropriate tools rather than paying FTEs to 'reinvent the wheel' is certainly a better use of company resources.

Pervasive has data connectors for just about anything, even really old 132 column report type stuff! (Only useful if you're dealing with old mainframe stuff)

I've used it to convert Progress to Oracle DW, Progress to SQL and vice-versa, Foxpro to SQL. On a smaller scale I used it to convert Quickbooks to SQL.

SSIS is certainly better than the old DTS but far behind 3rd party tools.

thumper_disc Valued Member Points: 62 More actions · Answer 9

I appreciated you article on SSIS. It is timeconsuming to use SSIS in a lot of situations like yours.

You might be able to eliminate SSIS completely. You can use xp_cmdshell to run operations like ftp directly on the operating system. You can also unzip. Just build the commands you want.

Use insert / exec to pull the directory contents into the stored procedure.

Use bulk insert to load data into a database table. The format file option is great for fixed length records and the fastest way I know to load data.

Steve Jones - SSC Editor SSC Guru Points: 734416 More actions · Answer 10

Martyn Hughes (10/19/2010)
My comment is not meant to be 'snarky', I'm always troubled when I have to fight within an organization to get useful tools. Pervasive does cost several thousand but the amount of time saved using appropriate tools rather than paying FTEs to 'reinvent the wheel' is certainly a better use of company resources.
...

Thanks, and I would agree with you. there are many tools that are worth buying and will help you do a better job with a good ROI. It's good to hear where other tools might fill in a space.

My comment was there as it appeared you were knocking this article as a way of accomplishing a task without providing any reasoning. Thanks for the clarification.

Geoff A SSChampion Points: 11511 More actions · Answer 11

Its good info. Gives me some new ideas on how to handle text files....

I have always used a User Defined Function to get text files into a table.

--------------------------------------------------

-- USAGE :

-- Select line from

-- Dbo.uftReadfileAsTable('MyPath','MyFileName')

--------------------------------------------------

USE [master]

GO

SET ANSI_NULLS ON

GO

SET QUOTED_IDENTIFIER ON

GO

Create FUNCTION [dbo].[uftReadfileAsTable]

(

@path VARCHAR(255),

@filename VARCHAR(100)

)

RETURNS

@file TABLE

(

[LineNo] int identity(1,1),

line varchar(8000))

AS

BEGIN

DECLARE @objFileSystem int,

@objTextStream int,

@objErrorObject int,

@strErrorMessage Varchar(1000),

@command varchar(1000),

@HR int,

@String VARCHAR(8000),

@YesOrNo INT

select @strErrorMessage='opening the File System Object'

EXECUTE @HR = sp_OACreate 'Scripting.FileSystemObject' , @objFileSystem OUT

if @HR=0 Select @objErrorObject=@objFileSystem, @strErrorMessage='Opening file "'+@path+'\'+@filename+'"',@command=@path+'\'+@filename

if @HR=0 execute @HR = sp_OAMethod @objFileSystem , 'OpenTextFile'

, @objTextStream OUT, @command,1,false,0--for reading, FormatASCII

WHILE @HR=0

BEGIN

if @HR=0 Select @objErrorObject=@objTextStream,

@strErrorMessage='finding out if there is more to read in "'+@filename+'"'

if @HR=0 execute @HR = sp_OAGetProperty @objTextStream, 'AtEndOfStream', @YesOrNo OUTPUT

IF @YesOrNo<>0 break

if @HR=0 Select @objErrorObject=@objTextStream,

@strErrorMessage='reading from the output file "'+@filename+'"'

if @HR=0 execute @HR = sp_OAMethod @objTextStream, 'Readline', @String OUTPUT

INSERT INTO @file(line) SELECT @String

END

if @HR=0 Select @objErrorObject=@objTextStream,

@strErrorMessage='closing the output file "'+@filename+'"'

if @HR=0 execute @HR = sp_OAMethod @objTextStream, 'Close'

if @HR<>0

begin

Declare

@source varchar(255),

@Description Varchar(255),

@Helpfile Varchar(255),

@HelpID int

EXECUTE sp_OAGetErrorInfo @objErrorObject,

@source output,@Description output,@Helpfile output,@HelpID output

Select @strErrorMessage='Error whilst '

+coalesce(@strErrorMessage,'doing something')

+', '+coalesce(@Description,'')

insert into @file(line) select @strErrorMessage

end

EXECUTE sp_OADestroy @objTextStream

-- Fill the table variable with the rows for your result set

RETURN

END

your method works well also. Thanks....

http://www.simple-talk.com/sql/t-sql-programming/reading-and-writing-files-in-sql-server-using-t-sql/

SQLRNNR SSC Guru Points: 281334 More actions · Answer 12

Thanks - I needed something like this.

Jason...AKA CirqueDeSQLeil
_______________________________________________
I have given a name to my pain...MCM SQL Server, MVP
SQL RNNR
Posting Performance Based Questions - Gail Shaw[/url]
Learn Extended Events

shresthadk Valued Member Points: 50 More actions · Answer 13

Hi Sarah

While testing I found that dbo.[Load_RNFile] is not defined anywhere in your script and am bit lost (or may be am totally wrong) Could you please review and profide the schema of dbo.[Load_RNFile]

Thanks

Sarah Old Hand Points: 352 More actions · Answer 14

The format for the table would look like this:

CREATE TABLE [dbo].[Load_RNFile](

[Id] [int] IDENTITY(1,1) NOT NULL,

[EverythingElse] [varchar](500) NULL,

CONSTRAINT [PK_Load_RN] PRIMARY KEY CLUSTERED

(

[Id] ASC

)

) ON [PRIMARY]

GO