Shortly after I became the system admin at lsu. We managed to spec
out a new machine for our ageing sparc 1000 server with 2 processors
and 512mb ram. Running oracle and what was then a 8 or 9 gigabyte
database.
Before the new machine was put in place our average load on the
system during the day as always pegged at greater than 20 and closer
to 30.
Payroll literally ran each payroll process, walked away and came back
a couple hours later and it still would be chugging along... on the
new server payroll managed to finish all their jobs in less time than
a single job used to take....
On May 13, 2007, at 10:09 AM, Roger E. Rustad, Jr. wrote:
You gotta love the loadaverage -> effect chart
LoadAverage Effect
0 New processes start and run quickly. The system is very responsive.
If this is a server, then you spent too much money on it, since it's
mostly sitting idle
0.5 WindowsNineX? machines begin to feel unresponsive.
1.0 The machine is processing incoming events quickly and
efficiently. Many machines running CPU-intensive background tasks sit
at this level for months at a time.
2.0 The machine is beginning to earn its keep. It is still running
efficiently, but some interactive tasks begin to feel "jittery."
4.0 Interactive tasks are noticeably jumpy.
6.0 The system may be beginning to thrash.
10.0 A FreeBSD server begins to wake up and wonder what the ruckus
is about.
15.0 Some older Unix systems begin to cry.
30.0 FreeBSD begins to cry, but keeps working.
40.0 That $300,000 Sun Enterprise Server begins to seem pretty
reasonable.
_______________________________________________
909linux mailing list
[email protected]
http://909linux.org/cgi-bin/mailman/listinfo/909linux
