On Tue, Aug 18, 2009 at 2:06 PM, Jesse Keating<jkeating(a)redhat.com> wrote:
The hit on slow disks is pretty bad too, even non-slow disks. With
a
local mirror the time to re-make the deltas is far longer than the time
to just pull down the entire packages.
Time isn't the only metric...people may also be concerned about
metered bandwidth. CPU overchurn we can limit via process
priority..but finding a heuristic to determine what "too slow" means
when it comes to disk io versus bandwidth is probably something we
can't do.
-jef"speaking of metered bandwidth..I need to remember to work from
barnes and nobles instead of working from home this winter..so i can
steal power, wireless and a nice warm fire"spaleta