Re: turning off 'atime' on zfs files systems for datafiles

  • From: Mladen Gogala <gogala.mladen@xxxxxxxxx>
  • To: oracle-l@xxxxxxxxxxxxx
  • Date: Mon, 12 Feb 2018 16:00:01 -0500

Hi Matt,

The "noatime" mount is an old performance trick, for all Unix like operating systems and for all file systems. There is no danger of using it. You will save yourself a considerable amount of IO, because the inode (file definition) is normally updated on access. When you specify "no access time" mount, the inode is updated on file close and not in the mean time. Of course, extending files will still require an inode update. Here is an article relevant to Linux. Slowaris is probably just the same.:

ftp://ftp.iitb.ac.in/LDP/en/solrhe/chap6sec73.html

Regards


On 02/12/2018 12:12 PM, Matt Adams wrote:


Has anyone out there investigated whether there is any performance impact to be gained by turning off the   “Update access time on read” feature of ZFS filesystems or if there is any danger in doing so when those file systems are used for holding datafiles or online redo log files?  (not using ASM)

We’re on RDBMS 11.2.0.4 on Solaris (sparc)

Matt

**** This communication may contain privileged and/or confidential information. If you are not the intended recipient, you are hereby notified that disclosing, copying, or distributing of the contents is strictly prohibited. If you have received this message in error, please contact the sender immediately and destroy any copies of this document. ****


--
Mladen Gogala
Database Consultant
Tel: (347) 321-1217

Other related posts: