Understanding NCHAR and Data Storage.

Question

Hi
I plan to migrate an Oracle8i database to Oracle9i to support UNICODE.
After reading Oracle9i Daily Feature at
http://otn.oracle.com/products/oracle9i/daily/oct02.html
I just don't understant how data would be stored.
e.g.: if I modify existing tables by altering columns from CHAR to NCHAR (or VARCHAR2 to NVARCHAR2), what happens to storage. Is it doubled ?
Thanks.

Answer 1

That depends on the character set(s) you use and the type of data you use.
Take the scenario where your current database character set is WEISO8859P1 and you start using UTF8 as the national character set. UTF8 is a variable-width encoding system, so it will use between 1 and 3 bytes to store each character. All english characters are encoded with a single byte, Western European characters umlats, accents, etc) would generally require two bytes, and Asian characters would require three bytes. If 99% of your data is just English characters, going from ISO-8859 to UTF8 wouldn't change the amount of space required to store your data in any significant way. If you started storing a lot of Chinese data, however, it would take three times the storage to hold a Chinese character than it did to store an English character.
Now lets imagine a scenario where you're going from an ISO-8859 encoding to a UTF-16 encoding. Since UTF-16 is a fixed-width encoding, it takes two bytes to store every character regardless of language. Character based columns in an existing ISO-8859 database would require twice as much disk if the encoding were changed to UTF-16, but Asian languages would only require two, rather than three, bytes per character.
If you're going to be storing a lot of Asian language data, the UTF-16 will end up requiring less physical space. If you're going to be storing mostly Engligh & Western European languages, UTF-8 will require less physical space.
Justin
Distributed Database Consulting, Inc.
www.ddbcinc.com/askDDBC