Another question about thrashing

There didn't seem to be much interest in my earlier post about kernel behavior when swap thrashing.

So my question is, are we not concerned about system behavior when one process uses so much memory that it grinds everything else to a halt?

It appears that a thrashing process is being given more preferential treatment than it should.

Design characteristic of global page replacement algorithms. It's not getting touched for 2.5/2.6

-- wli