Charting server load
Zbigniew Koziol
softquake-Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org
Thu Feb 1 21:16:30 UTC 2007
On Thursday 01 February 2007 15:18, Lennart Sorensen wrote:
> On Fri, Feb 02, 2007 at 01:21:30PM +1859, Ian Petersen wrote:
>
> The load is simply: On average how many processes are in the 'ready to
> run' state. If on average, 0.5 processes are ready to run, then you can
> only be using 50% cpu, while if 3 processes are, then you must be
> running at 100% cpu, unless you have more than 3 processors.
Not quite true.
In case of Linux, it is not only the number of processes that counts but also
the number of attempts to read or write to a file (writing in particular is
time consuming).
Hence, CPU might be using 5% of its ability, RAM might be used in 10% only,
while load might go to 50 and the computer will slow down tremendously when
for instance a heavy usage of database takes place. Adding CPU or RAM will...
surprise, surprise! will NOT help! The system spends simply too much time in
a state when attempts to get to hard drive are done.
I observed that situation when experimenting with postgres. Solution which is
used in similar cases is... putting postgres data to RAM instead on HD.
zb.
--
The Toronto Linux Users Group. Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists
More information about the Legacy
mailing list