FLAT FILE ON UNIX BOX TO ANY RDBMS DATABASE

[PHOBY]

I am having a requirement as follows...

I have one flat file on unix box contains 150K records line by line. I need to pick up & read these files and should transform (or massaging) with some buiness rules and should put these values in one RDBMS. Application server will work in cluster environment and powerful.

I am looking for a strategy which should satisfy me for ...
1. Application server can be IBM websphere and Web logic
2. RDBMS can be Oracle or SQL Server or UDB DB2
3. Process should be initiated using webservice
4. All the business logic should be in JAVA
5. Database can be changed in future. it means now I will use Oracle and ater one year I may change it to SQL server. So no code change is allowed. Stored procedure is not allowed.
6. Time window available is 2 hours.
7. business transformation usecases should be portable,reusable & deployable such as JAVA.

So, is there any casestudy or white paper or any high level strategy can be suggested by anybody?

Please help me on this. This is very urgent.