I've been spelunking around the VM system as my kernel profiling seems to
indicate that it's quite inefficient -- which will likely come as no surprise
to many. :-)

Looking through vm_pageout.c, I wonder why CLUSTERED_PAGEOUT is turned off.  I
don't have any machines that really page enough for this to be an issue, but
it seems like it would be a fairly substantial win if it were turned on.

Is there some reason that this is turned off?  Support for it appears to exist
in the pagers, and in any event there's a per-pager interface to indicate
whether or not it's supported.

Thor