>David
> What types of operations are you performing on the data once it is on the client machine ?
>
>Are you using something like a Fill() method to fill your database.
>
>If so I might recommend using a data reader which will process through each record one at a time. Readers are one way mechanisms that only read forward. You do have access to each column of data via this mechanism but you cannot go backwards.
>
>We had some rather large SQL to XML exports to do and by using a data reader we were able to create a 1.1GB file in 5 minutes. We had comparable amounts of records and a very large record structure.
>
>Rodman

We are, deep in the bowels of our business object, using Fill().

I wondered if putting the file in a format that prevented a lot of swapping in and out of RAM would help the problem. By going through a file unidirectionally, I thought it might prevent a lot of swapping if the data was only handled in small chunks. Do you think this works because using a data reader requires less RAM swapping?

Any guess as to the size of a DataTable stored in virtual memory on hard disk compared to the same data as XML? I haven't a clue.

Thanks for your suggestion.