Database caching?

On further reflection, there IS a better way to do this.  It only makes sense because you have such a huge, huge amount of data to move. 

If you have two SQL Server databases running on the same server, you can write a stored procedure where one database selects data from the other database.  I don't remember how to do it.  I haven't got time to look it up right now, but it's possible.  I believe it's possible in DB2 as well.  I don't know about other databases.  Doing it this way may have performance benefits.

As with ADO.net, you might want to write a stored procedure that gets the data a bit at a time.  For example, it might keep selecting 5000 rows at a time until there are no more rows.

As an aside, it still seems unusual to have to move so much data on the fly.   It should be avoided altogether if possible.

There are ALWAYS alternatives.  In this case, however, why would you want one?  If you're making a .Net application, you should stick with .Net unless there's some compelling reason to find another way.  Just in terms of making something easy to maintain, it makes sense to stick with the same technology throughout.

Your real issue here is that you're trying to move a HUGE amount of data on the fly.  A million records is a lot of data to copy.  It's processor intensive.  It could cause transaction timeouts if you're not careful.  It makes sense to break it into several smaller updates.

Yup.. I perfectly get that point mate.. But I need to save the intermediate data (till my failure point) right (thats 0.1 million rows).. How do I do that... 
Now let me put this as a combination of 2 small processes..
1.  I will intermediately save the 0.1 million (thats till my failure point)
2.  I will restart my process from the 0.1 million instead of the entire contents..
My question is more related to point 1, how will i code in such a manner so that the intermediate data gets saved.

So you want to restart after failure, correct?
 
You can get the maximum value for the key in the output table, then select all records in the input table with a key greater than that value.

Apologies if this is confusing.  I will try to simplify things.  I have an input table, Table A containing 1 million rows.  User will use my application to import the contents of Table A into the app DB, lets say Table B.  Internally I am using ADO.net to do this data transfer.
If data retrieval from Table A fails or the whole process stops at 0.1 million at the input stage, What will be my possible alternatives.
1.  Import Table A again into my app DB (unnecessary overhead)
2.  My app saves till 0.1 million (i.e) till the failure occured.  Then I will import Table A again but this time i will move the rest of the 0.9 million rows only instead of the entire 1 million.
I like option 2, and hence my question is how to code in such a manner that the data till the failure point is saved in my App DB.

Yeah.  I do have the primary key sorted out...  I guess you answered the second part of my question which is resuming from the point it got failed.  But first problem is I need to have the data transferred to my target database till the failover point.
To make my question more clear, 
I am using ado.net data adapter to fill out a dataset lets say from table A, which contains 1 million rows.  Once it fills out, I am saving this initial copy somewhere in my app database.  If some failure occurs at around 0.1 million records, how do i move the 0.1 million rows that has already been transferred  to the dataset to move into my app database, so that i can resume from the point it failed.
Regards
Raja