Regarding best option to load to oracle database

[dsdev_123]

Hello

I have a requirement like i have a huge file with 10 million records and i need to load the data to oracle table. We are using Parallel Edition on unix AIX box. Which option would be the best.I also have a requirement where if the job aborts in the middle of loading i need to rollback on what ever records got inserted/updated.

Appreciate your help

Thanks

[Sainath.Srinivasan]

Not enough info.

It is like asking "how fast can a person with two legs run?"

Try it with a sample and tune variables.

[chulett]

Well... your rollback requirement means no direct path bulk loads. So you'll need to stick with conventional loads but we'd need more information to give better advice - always inserts? A mixture of inserts / updates? If a mixture, any idea about the ratio?