From: Evans, Steve (steve.evans@Fiserv.com)
Date: Wed Apr 12 2006 - 14:51:12 EDT
Hello admins,
I'm new at a company and looking at performance issues. Most of the
problems are application/ coding related, but there is one metric I am
tracking that is giving me fits and may be system related.
GBL_MEM_PAGE_REQUEST tracks the total page ins and page outs over the
interval (as vmstat -s reports). The interval I'm looking at is 5 minutes.
It jumps up and down around 1.5M, then at the same time every day it
plateaus at 3M, which is ~1000 requests/ sec. There are several cron
entries that could be the culprits, but to avoid 'tainting' your answers,
I'll refrain from listing those, because it lasted for most of a day during
one week.
Q: Why is the total number of pages plateauing at 1000/sec?
I am used to seeing metrics spike, not plateau. Is there a hard physical
limit or kernel parameter at work here?
Graphical, high tech example of what I'm looking at:
__ ______
| | / |
./\| |/\/\/ |
Thanks,
Steve Evans
UNIX Systems Administrator
-- ---> Please post QUESTIONS and SUMMARIES only!! <--- To subscribe/unsubscribe to this list, contact majordomo@dutchworks.nl Name: hpux-admin@dutchworks.nl Owner: owner-hpux-admin@dutchworks.nl Archives: ftp.dutchworks.nl:/pub/digests/hpux-admin (FTP, browse only) http://www.dutchworks.nl/htbin/hpsysadmin (Web, browse & search)
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 11:02:52 EDT