DSXchange

Friends,

Currently I'm working on designing a job that transfers data between 2 Oracle databases. Both the databases are set to UTF8 NLS character set. I could clearly see the Japanese characters (す下る)when I login to the source db.

The problem is that the Japanese characters are being replaced by "????" in the destination db when I execute the DataStage job that transfers the data.

The design of the job is very simple
ODBC ==>Transformer==>ODBC

Am I doing something wrong

Thanks in advance
Yamini

Is UTF 8 characte set is enabled at job level?
Try to remove the ustring (entended property of meta data).

UTF8 is set as the default NLS at the project level. Also, can you please tell me how I can remove the ustring?

Thanks
Yamini

If you remove the ustring you definitely will not be able to process Japanese characters. However, if you wish to do so, click on it - it's a drop down list in which the other option is empty. Choose that option.

Ray,

We do want to process the Japanese characters. Can you please let me know the correct way to process these characters while populating data in the destination table

Thanks
Yamini

You need to obtain a definitive statement concerning how these characters are encoded in the target table (for example SHIFT-JIS, UTF-8), then ensure that all DataStage mappings and all database mappings are compatible. For example if your target is Oracle, make sure that the NLS_LANG environment variable is correctly set (check the "environment variables" entry in the job log).

Ray,

Data in the source is encoded in UTF-8 format. I have changed the NLS at the project level to UTF8. Also, added an environment variable NLS_LANG with "AMERICAN_AMERICA.AL32UTF8" as the value.

I'm still not able to process the Japanese characters. Can you please point me to the right direction

Thanks
Yamini

I finally figured out the problem. All the columns containing Japanese characters were listed as "VARCHAR". The job started to process the characters as soon as I changed the definition of the columns to "NVARCHAR"

Well done. I guess we'd assumed you'd used NVarChar.