![]() | |
![]() |
| | Thread Tools | Display Modes |
#1
| |||
| |||
|
|
Hi all When using DTS importing data from my AS/400 - DB2 data base was a trivial matter, all I did was select client access 400 ODBC then select my tables and all was done. I cannot find anything like this in SSIS. the ODBC connection does not list the tables in the library thus forces the entry of a SQL statement then it incorrectly detects the data type of the cols. so I have to manually change the data types for all the cols. The OLE DB connection detects 3 tables in the library ( there are over 200) and we are back again to the ODBC issues I really want to use SSIS but this is making it extremely painful (not to mention long winded ) to do. Can somebody help me with making this easier Thanks Reg |
#2
| |||
| |||
|
|
Reg Besseling wrote: Hi all When using DTS importing data from my AS/400 - DB2 data base was a trivial matter, all I did was select client access 400 ODBC then select my tables and all was done. I cannot find anything like this in SSIS. the ODBC connection does not list the tables in the library thus forces the entry of a SQL statement then it incorrectly detects the data type of the cols. so I have to manually change the data types for all the cols. The OLE DB connection detects 3 tables in the library ( there are over 200) and we are back again to the ODBC issues I really want to use SSIS but this is making it extremely painful (not to mention long winded ) to do. Can somebody help me with making this easier Thanks Reg I agree with using the OLEDB provider from Microsoft. That solved many problems for us also. In our experience it is not the use of a SQL Statement that provides all of the Field type conversion problems. We get that whether we use a SQL statement or specify a table. In my opinion it is a bug in the SSIS or Visual Studio environment. What we found solves all of those problem is to ALWAYS use the wizard to create the package. Nearly all packages we build using the wizard have zero problems with data types. All of those that we created manually forced us to put an explicit conversion into the Data Conversion Task which is really a PITA on large tables. We even find that if we create a package with the wizard and then need to modify it to include fewer or more fields that we have to use the wizard to build the new task from scratch. Any attempt to modify the existing package again forces us to explicitly add conversions to nearly all text fields. -- Rick Brandt, Microsoft Access MVP Email (as appropriate) to... RBrandt at Hunter dot com |
#3
| |||
| |||
|
|
Rick, Have you ever compared the Wizard generated package to a hand built to detect the differences. I see no reason why you could not build the same package by hand, do you think otherwise? Just interested. Saying that using the Wizard to do some of the hard work seems like a no brainier. In case you are not aware, the Wizard uses mapping files to help it make accurate data type selection between different provider types, and you are obviously getting the benefit of this and quite like it. It is a shame some of the source components do not consume the same mapping files when used directly in the designer. |
#4
| |||
| |||
|
|
Try this with both a wizard in DTS and a wizard in SSIS: Crate a package that exports data from a SQL table and dumps into an Excel spreadsheet. Create a second package that takes from that same spreadsheet and puts it back in the same SQL table. Accept all the defaults It's a no-brainer for DTS. The SSIS Wizard gets bound up in unicode vs non-unicode conversions on the second package. Took me three days to troubleshoot that one. Now THAT's a PITA. I don't use any wizards anymore. I use my own Package templates. |
![]() |
| Thread Tools | |
| Display Modes | |
| |