I inherited a package which creates a flat-file on a remote pc ( ex.
\\remotepc\dir1\file.txt). When the output file appears in the remote
directory, it grows very slowly. The sql which is the DTS data source is a
normal sql statement with a Group By. I take this to mean that the entire
set of data is retrieved and grouped before the file starts writing. Is this
true? Would it be faster to create the file "locally" and then copy the file
to the remote machine?

Michael

That would be how I would do it. I would bulk copy the data to a file
first. Then I would use the built in "File Transfer Protocol" task to
transfer the file as the next step, or even xcopy the file with
"Execute Process" task.
In fact all this could be done as steps in a SQL Server Agent job. I
tend to do as much as I can there because the logs are easier to read
and I can restart from a particular step more easily.