From: joe_fletcher@btconnect.com
Date: Fri May 28 2004 - 09:14:07 EDT
Hi,
Hoping someone might be able to give me a few ideas on this.
We've got a system running oracle being used for performance testing. The
tests are
run from a baseline image of the database. Compressed copies of the dbf
files are help
on disk and restored at the end of each test to regenerate the baseline
state.
During normal operation the (veritas) volumes are mounted using the
following options.
rw,suid,log,largefiles,mincache=direct,convosync=direct,noatime
However when they run the uncompress/restore it takes 5 hours. If the
volume are
dismounted then remounted without directio enable thus:
rw,suid,log,largefiles
then the restore runs in under an hour. The restore uses standard unix
compress.
The problem is the fact that the volumes need to be remounted with the
different
options each time they do a restore then put back for the next test. This
involves
sysadmin interactions ie root access. We'd like to remove this requirement.
I have two options. One is just to write an RBAC script to allow the apps
users to do the
job themselves (we aren't about to hand out root access). The other is to
find out why
the restore takes so long with directio enabled and if possible find a
solution that
doesn't run slowly using directio.
Base system is a F15K domain running Solaris 8 talking to EMC Symmetrix
storage.
Any ideas?
TIA
Joe
_______________________________________________
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 - 23:28:44 EDT