F
frank.meulblok
Guest
Speaking of I/O, I'd suggest testing with an .st file with fixed extents. I've been testing something similar with type 2 areas and the I/O overhead to grow the variable extents was enough to be a real bottleneck. Setting up the .st file to use fixed extents shaved a few seconds off of loading the .df. (With potentially much bigger gains to be had if you also need to load actual data.) For type 1 areas the difference isn't going to be as big (for all the reasons Rob already pointed out above), but I'd still expect to see some gain.
Continue reading...
Continue reading...