I have done this in the past, and this is not a trivial task. We ended up writing each table to a flat file with channel separators and reloading each table in Oracle using Oracle SQL Loader. There was a ton of Perl scripts to clean up the source data and shell scripts to automate the process as much as possible and run operations in parallel.
Gotchas that may arise: 1. Select a delimiter that is as unique as possible. 2. Try to find data types that are as close to Informix as possible. those. date and time stamp 3. Try to get the data as clean as possible before flushing flat files. 4. HS is likely to be too slow.
This was done many years ago. You might want to explore the Golden Gate software (now owned by Oracle) that might help in this process (GG did not exist when I did this)
Another idea is to use the ETL tool to read Informix and upload the data to Oracle (Informatica comes to mind)
Good luck :)
source share