What you are seeing is the result of a badly configured
paging subsystem

What happens: Linux touches all it's pages when it boots.
These pages then overtime get paged out. Then you run your
program - and all those pages get paged back in.

Applications with more consistent workings sets would not
see this.

Erm... far be it from me to argue with the master on the performance of the actual physical hardware, but isn't the point here that they've generated a worst-possible-case scenario (is this called a degenerate case these days?) and that if everything needs to be paged back in then the problem is that they've inherently produced a paging scenario in their program? (Assuming that they've been paged out, of course.)

I know that the paging and pre-fetch algorithms on VM are
pretty good but isn't the best solution to this case to
change the program? Standard program performance and tuning?
Or even system tuning for any pre-fetch algorithms?

(Aside: I was surprised how much of my VM perf/tun work
carried over to Solaris when I went on their perf/tun course
and there's 15+ years between the two courses. Of course, I
was left wondering how people manage when one of the other
sysadmins did't want to know that it really still matters
where you put your swap, system utils etc. on a disk
because in the 21st century you didn't want to have to
be bothered with that. Sigh. Oink, oink, flap, flap.)

Rod

Reply via email to