R vs. S-PLUS vs. SAS

Hoon, 

Your questions were addressed to David, but I hope he won't mind if I
interject. "How to handle a large dataset" is practically a FAQ on the
R-help list: search through the archives at
http://maths.newcastle.edu.au/~rking/R/. To summarise, first it must be
noted that thoughtful use of R (scan(), avoid silently copying data in
memory, etc.) helps handle very large datasets reasonably quickly; the
limit is basically the available amount of RAM. If you need to work on
more data than that, the best practice is to put it in a database, and
to access it in segments. The R gurus also usually encourage the
research to think whether all that data is truly necessary, or if a
subset of it could be used to draw the same conclusions without too much
trouble. 

Cheers, 

Pijus