On optimising the scheduler for large run queues

[Jan-Simon Pendry]

Jamie Lokier wrote:
> [lots of good stuff]

i think this is one of those philosophical areas where a decision
one way or another needs to be made.

other commercial unix systems (hp, sun, ibm, sequent) tend to split
into two camps as follows:

1. those that optimise for performance, but degrade badly under intense
   load.

2. those that optimise for intense load, but don't have top-notch
   performance under low-load.

now, my bias is towards #2, because i am more interested in server
systems, and from a sizing, tuning and sysadmin perspective they
are easier to plan and manage.  however, this is probably not the
right thing for a desktop operating system.

so, which way is linux going?  you tell me - there doesn't seem to
be a "right" answer here.  this may well mean there needs to be two
schedulers, one with ultra-low overhead for desktops, and another,
with better scalability, for servers.

jan-simon.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo en vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/