Convert NLS_CHARACTERSET 9i RAC DB from WE8ISO8859P1 to UTF8

Question

Hi,
Currently my database has
NLS_CHARACTERSET=WE8ISO8859P1
NLS_NCHAR_CHARACTERSET=AL16UTF16
However, developers request to change NLS_CHARACTERSET to be UTF8 because new source of one of new data marts is UTF8. Currently there is issue with displaying only few records. For conversion, I need to recreate a new database which is what I'm trying to avoid.
My quesitons are
1. Since Unicode data stored in few columns in few tables in this new datamart only, can they take advantage of NCHAR and NVARCHAR instead?
2. If converted to UTF8, what is the drawback of storing all data in database in Unicode?
3. Is AL32UTF8 better (and superset) than UTF8?
Thanks

Answer 1

Generally, you won't be able to convert the existing database to UTF8 since UTF8 isn't a strict binary superset of the existing ISO 8859-1 character set. You would probably have to create a new database and export & import the data over. If there are special circumstances, you may have some other options, but it will involve quite a bit of administration.
Working with NCHAR columns tends to be a bit more complex from an application development perspective (though this complexity differs depending on the front end used). Depending on the language of most of the data, there are probably space implications of UTF16 vs UTF8. UTF16 is a fixed width character set, so every character occupies two bytes of storage. UTF8 is a variable width character set, so English characters will tend to be 1 byte, European characters will tend to be 2 bytes, and Asian characters will tend to be 3 bytes.
Justin
Distributed Database Consulting, Inc.
http://www.ddbcinc.com/askDDBC