Your volume size is large so don't discount that factor in everything you do. I don't know what your overall data size is, but I would imagine it is up around a terabyte or so. Much advice you will read doesn't apply to those implementations with large volumes so doing a vanilla implementation won't work.
Cloning is problematic and never recommended - not even for small-scale warehouses. Moving changed-only data is always recommended. You should extract from Siebel ONLY records that have changed since the last Siebel extract. This will take a fraction of the 17 hours and is more standard. By moving changed-only data at whatever interval you can - perhaps daily - you begin the process of moving towards more frequent loads. You could then perhaps run that ETL cycle intra-day and more frequently until it meets your real-time requirements.
Since you have said that the ETL cycle time is the issue, not query performance, I don't know if you have query performance problems related to your Business Objects setup.
You move the data to a DBMS, not "to" Business Objects. If you are having query performance problems, the vast majority of those are solved with DBMS (and below - operating system, hardware enviroment, etc.) tuning, not Business Objects tuning.
This was first published in December 2004