I have a non-utf8 oracle database running on 11.1.0.7. We need to support greek characters. So we have two options:

1. use nvarchar, nclob fields for those fields that need greek (it is not all fields). We have tested this and gotten it to work with java coding.
2. convert Oracle to AL32UTF8 database. I am not asking how to do this. I got this from the Oracle Site/Oracle Support. I know what is involved, lossy data, etc, increasing the size of the database.

My question is we have users to our system that connect to our database with database links but work on SQL Server and IBM DB2 databases. I do not have access to those databases and I do not have experience with them. If they are not in UTF-8 databases what happens when they pull UTF8 data?

I would assume that English/Ascii characters are fine and the greek will end up as junk data.

I also ran Oracle Character set scanner (oracle command line utility you use to get info about the affects of a character set conversion). It says that my database will crease in sizez by about 20%. Does this have an affect on users with 3rd party databases?

These are customers of our data and there is a limit to how much access I can have to them to run tests. Any information you have would be welcome.