PowerCenter Code Page Conversion
When in data movement mode is set to Unicode, the PowerCenter Client accepts input in any language and converts it to UCS-2. The PowerCenter Integration Service converts source data to UCS-2 before processing and converts the processed data from UCS-2 to the target code page before loading.
When you run a session, the PowerCenter Integration Service converts source, target, and lookup queries from the PowerCenter repository code page to the source, target, or lookup code page. The PowerCenter Integration Service also converts the name and call text of stored procedures from the PowerCenter repository code page to the stored procedure database code page.
At run time, the PowerCenter Integration Service verifies that it can convert the following queries and procedure text from the PowerCenter repository code page without data loss:
- •Source query. Must convert to source database code page.
- •Lookup query. Must convert to lookup database code page.
- •Target SQL query. Must convert to target database code page.
- •Name and call text of stored procedures. Must convert to stored procedure database code page.
Choosing Characters for PowerCenter Repository Metadata
You can use any character in the PowerCenter repository code page when inputting PowerCenter repository metadata. If the PowerCenter repository uses UTF-8, you can input any Unicode character. For example, you can store German, Japanese, and English metadata in a UTF-8 enabled PowerCenter repository. However, you must ensure that the PowerCenter Integration Service can successfully perform SQL transactions with source, target, lookup, and stored procedure databases. You must also ensure that the PowerCenter Integration Service can read from source and lookup files and write to target and lookup files. Therefore, when you run a session, you must ensure that the PowerCenter repository metadata characters are encoded in the source, target, lookup, and stored procedure code pages.
Example
The PowerCenter Integration Service, PowerCenter repository, and PowerCenter Client use the ISO 8859-1 Latin1 code page, and the source database contains Japanese data encoded using the Shift-JIS code page. Each code page contains characters not encoded in the other. Using characters other than 7-bit ASCII for the PowerCenter repository and source database metadata can cause the sessions to fail or load no rows to the target in the following situations:
- •You create a mapping that contains a string literal with characters specific to the German language range of ISO 8859-1 in a query. The source database may reject the query or return inconsistent results.
- •You use the PowerCenter Client to generate SQL queries containing characters specific to the German language range of ISO 8859-1. The source database cannot convert the German-specific characters from the ISO 8859-1 code page into the Shift-JIS code page.
- •The source database has a table name that contains Japanese characters. The PowerCenter Designer cannot convert the Japanese characters from the source database code page to the PowerCenter Client code page. Instead, the PowerCenter Designer imports the Japanese characters as question marks (?), changing the name of the table. The PowerCenter Repository Service saves the source table name in the PowerCenter repository as question marks. If the PowerCenter Integration Service sends a query to the source database using the changed table name, the source database cannot find the correct table, and returns no rows or an error to the PowerCenter Integration Service, causing the session to fail.
Because the US-ASCII code page is a subset of both the ISO 8859-1 and Shift-JIS code pages, you can avoid these data inconsistencies if you use 7-bit ASCII characters for all of your metadata.