SQL Server 2000 Collation Changer

Introduction

Do you have an SQL application that you need to deploy in another country? Do you need to change the collation of your SQL database and all objects in it? If the answer is yes, then this could be a very slow process to do manually. If you run the alter database script as below:

ALTER DATABASE [My_Database] COLLATE My_Collation

You will find that the database default collation is changed for new columns, but all existing columns will retain the original collation order. Changing the collation on each column is a non-trivial task, as you need to drop all indexes, full text indexes and constraints associated with the column, along with any user-defined functions. Once the collation order has been changed, you can recreate the indexes, constraints and functions. This C# tool simplifies the process by creating an SQL script that does everything for you.

Using the Code

Simply run the program, select your SQL Server database and a new collation order. You have two options: you can either simply get the program to create an SQL script that you can run later (press the "Script Only" button) or you can actually make the changes (press the "Script and Execute" button). Things to note:

Always back up your database before running this tool. I cannot guarantee that you will not lose data. A number of the statements cannot be run in a transaction, so there is no way of detecting a failure and rolling back.
On SQL Server 2000, nText columns will be recreated so your column order will be slightly different.
To run the script, the program sets the database into single user mode to run the ALTER DATABASE [db_name] COLLATE [Collation Name] statement. You should therefore ensure that there are no open connections on the database before running the script (use the stored procedure SP_WHO to identify any open connections).
Since the script will drop and then rebuild all indexes and foreign keys in the database, you will find that it could take a long time to complete (possibly hours).
All columns will be changed to the new collation order, even if they have a non-default collation before running the script.
SQL 2005 support is gradually being added. If you encounter issues or missing functionality, please let me know.
If you change from a case-insensitive to a case-sensitive collation order, you may find errors occurring when recreating check constraints and functions. This is because your scripts will be parsed in a case-sensitive manner once the collation order has been changed. To work around this, I would recommend running the script in the program and reviewing the output once the script has run to completion. The error messages relating to each failure will be displayed in red under the code that failed.

Change History

18 January 2006

Original posting.

7 March 2006

Fixed bug where scripting failed on databases with case-sensitive collation orders. Script no longer drops foreign key constraints unless necessary.

30 August 2006

Fixed bug when scripting objects owned by different owners than the owner executing the script are used. May be seen as an error when scripting to #spindtab_____.
Fixed bug where scripting did not recreate permissions on table functions after they were recreated.

20 March 2007

Triggers now disabled while changes are made.
Altered sequence of execution to prevent errors following user feedback.
Added some SQL 2005 support. Let me know if you encounter any issues.
Resolved issue when recreating a table function where the body of a table function is greater than 4000 characters.
Reinstated the functionality to only delete the required indexes and primary keys.

10 October 2007

Added support for changing collation where full text indexes exist, including changing language used in the full text search.
Fixed issue when insteadof triggers exist on views.
Fixed issue where nText columns end up allowing nulls when they were originally declared NOT NULL.
SQL 2005 allows collation order of nText columns to be modified. The script has been modified to reflect this new functionality.
Fixed issues encountered when user-defined data types are used.
Added additional SQL 2005 support, including refactoring scripting logic to allow various parts of the script to be customised for different versions of SQL server. If you need to debug or customise the scripts, it should be a little easier to understand.
Converted from VB.NET to C#, as I use C# at work all the time and it is more familiar to me.
Moved execution task to a worker thread to give a more responsive UI.
Fixed issues when system databases have case-sensitive sort order.

1 March 2008

Created new code for handling indexes, statistics and relationships in SQL 2005. This new code uses the 2005 schema views and adds scripting support for new SQL 2005 functionality, such as included columns in non-clustered indexes.
Fixed issues where ANSI_NULLS setting is not correct after recreation of table functions.
Various minor UI bugs fixed.