Your first suggestion was: "Any way you could extract the textual
content and store it in
the db?"

We could do this, but we thought it might be better to store the binary
in the database and not have a copy of the binary on the file system
with a copy of the full text in the database and a pointer in the
database to the binary file as well as a mechanism to extract the full
text from the file.  What's your take on this?

Also, even inserting the text instead of the binary, if the file is
sufficiently large I can imagine a situation in which inserting the
blob the standard way could cause some very un-needed hard drive
thrashing.  (The standard way being reading all the text into a
variable and then supplying that variable as a parameter to an ad hoc
SQL query or Stored Procedure.)  In ADODB there was the Stream object,
but there doesn't seem to be an equivalent in ADO.NET.  Do you have a
suggestion for a good way to go about this?

Finally, we also wanted to return a sample of the text that matched the
search with our results.  I know that this is not possible in SQL
Server, so I'm wondering there might be an entirely different approach
for both storing and searching these documents that we should consider.

Thanks for all your help and feedback!
Kevin