|
[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]
|
[PLUG] Survey says: noatime?
|
A co-worker and I have been looking into the area of system performance
tuning, specificly as it relates to databases. We've come across a
filesystem technique mentioned in some of the articles we found:
http://www-106.ibm.com/developerworks/library/l-fs2.html
http://www.linuxjournal.com/article.php?sid=5840
The technique is to mount the file systems with the noatime option. Trying
out a simple little query in postgres without noatime on reiserfs we had
a query time of 6.31s, after adding noatime (and rebooting, I tried just
-o remount,noatime but it didn't seem to take effect untill the reboot) the
time for the same query dropped to 2.09s. I was surprised that it was such
a dramatic difference.
My question to the list is, am I hurting myself by not having the last
access time updated? I know how to look at a file's atime (-A in perl, or
ls -u) but I don't know of anything that uses or relies on it. What is
atime typicly used for? Is this something I can just disable for all of
my file systems to get a performance boost? If it's such a desierable
thing to do with very little drawbacks, why isn't it a standard option?
The articles mention using chattr +A to set it for specific files or
directories, but it seems that reiserfs doesn't support chattr. What
kinds of experiences do you have with journaling file systems and
performance?
Thanks,
Kyle
--
------------------------------------------------------------------------------
Wisdom and Compassion are inseparable.
-- Christmas Humphreys
mortis@voicenet.com http://www.voicenet.com/~mortis
------------------------------------------------------------------------------
______________________________________________________________________
Philadelphia Linux Users Group - http://www.phillylinux.org
Announcements-http://lists.phillylinux.org/mail/listinfo/plug-announce
General Discussion - http://lists.phillylinux.org/mail/listinfo/plug
|
|