From: Filipe Litaiff (filipe_litaiff@optiglobe.com.br)
Date: Wed Apr 07 2004 - 11:21:58 EDT
Hi folks, sorry for the late summary.
Special thanks to
Jose Luiz
Darren Dunham
Roy Erickson
Conclusions:
- The consensus is that regardless possibly some I/O issues, the process is
pretty harmless, as it uses the intent log on the final steps.
- I have shceduled the defrags on allfilesystems (200 Gb) and in some cases,
It took all my backup window (9 hours!) , where the applications are
unavailable, and eventually got through work hours. There is a feature what
limits the amount of time the defrag can run, probably I4ll use it.
- I have analyzed the IO on the hours the reorganization runned, and figured
that as expected, the I/O overhead is measurable but doesn4t seem to be
significant. Anyway, I choosed to schedule in a way tha it will never colide
to full backups, when IO is demanded heavily.
- It souds pretty on the manuals, but some feedback I had is that the
improvement it could deliver is hard to measure, or even non significant in
most cases. I undestood that it is caused by the very robustness and
flexibility of the VXFS itself, thanks to the good job Veritas did on its
filesystem. Anyway, I4ll try to figure out if someone here perceived
enhancements on performance.
Regards,
Filipe Litaiff
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 23:28:25 EDT