Check for growth of the os filesystem cache. The sysinternals utility
Rammap.exe may be of assistance. Not only does it map memory usage but
starting up multiple instances of it in series is a great way to wrestle
memory away from the "standby" pool of the is filesystem cache (memory
freed after the programs are terminated).
Leverage the use of xcopy /j and robocopy /j as well as RMAN to avoid
leveraging buffered io.
Paul
On Sep 30, 2015 3:10 PM, "Rodrigo Mufalani" <rodrigo@xxxxxxxxxxxxxxx> wrote:
Hello,
In addition we are running these logical backup in a ReFS (Micro$oft),
but this device was delivery to us a long time ago (more than year) and
started issue about one month ago.
All the best,
Rodrigo Mufalani
Database Consultant
+55 XX(21) 3193-0326
rodrigo@xxxxxxxxxxxxxxx
www.mufalani.com.br
"Stop guessing… Be Sure… Because performance matters"
Em 30/09/2015, à(s) 09:59, Rodrigo Mufalani <rodrigo@xxxxxxxxxxxxxxx>
escreveu:
Hi Michael,
Yes, I was wondering about that bug, I’ve increased my streams_pool_size
to 250mb and found in a german blog a simple solution, at least I don’t
faced to the event "AQ: enqueue blocked on low memory” anymore, but
still slow.
The workaround for that bug that you pointed.
http://www.dbmasters.at/db/masters/artikel/expdp-impdp-haengen-lange-zeit-und-sind-dann-sehr-langsam
SQL> alter system set aq_tm_processes=2;
All the best,
Rodrigo Mufalani
Database Consultant
+55 XX(21) 3193-0326
rodrigo@xxxxxxxxxxxxxxx
www.mufalani.com.br
"Stop guessing… Be Sure… Because performance matters"
Em 30/09/2015, à(s) 09:52, Michael McMullen <ganstadba@xxxxxxxxxxx>
escreveu:
have you tried explicitly setting streams_pool_size? google
streams_pool_size expdp or check out metalink.
------------------------------
From: rodrigo@xxxxxxxxxxxxxxx
Subject: expdp and exp are slowing (iscsi)
Date: Tue, 29 Sep 2015 18:42:47 -0300
To: oracle-l@xxxxxxxxxxxxx
Hi Folks,
Our env:
Oracle 11.2.0.4 SE (64bits)
Windows 2012 (64bits)
ISCSI
About one month ago one of our database servers, running on windows 2012
as I said on top, started an issue on logical backups (I tried expdp and
exp) both of them are slowing generating dumps after 3 or 4 days after a
boot on server. If we bounce only the database and listeners, the
performance still poor. When we restart the entire OS, backups are
generated with normal time, about 3 hours for a 1TB database, I think it is
good.
I’ve traced the backup sessions, a lot time on CPU and no events
indicating I/O issues.
My SysAdmin ran some storage performance softwares and haven’t issues
evidenced. The users are not complaining about performance issues, only the
backups, event the RMAN backup still running ok on same device.
Any directions will be appreciated.
All the best,
Rodrigo Mufalani
Database Consultant
+55 XX(21) 3193-0326
rodrigo@xxxxxxxxxxxxxxx
www.mufalani.com.br
"Stop guessing… Be Sure… Because performance matters"
