RE: ASSM Performance Problems

From: Jesse, Rich <Rich.Jesse_at_quadtechworld.com>
Date: Thu, 23 Jun 2005 14:28:11 -0500
Message-ID: <FB5D3CCFCECC2948B5DCF4CABDBE66971FFA4B@QTEX1.qg.com>

<whew> For a second there, I thought this was important. Looking at our ERP system running 9.2.0.5 on HPUX, I see only 14 affected tablespaces.  

Wait.  

That's all of 'em, save for SYSTEM, UNDO, and TEMP.  

Aw crap.  

Time to patch! At least we're not using bitmapped indexes, which seems to be the most popular culprit...  

Rich

	-----Original Message-----
	From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Robyn
	Sent: Thursday, June 23, 2005 1:56 PM
	To: jrsmiley_at_gmail.com
	Cc: oracle-l_at_freelists.org
	Subject: Re: ASSM Performance Problems
	
	
	John,
	
	Per Note:281047.1, use of ASSM with locally managed tablespaces

can cause index corruption in 9.2.0.5 RDBMS. There is also a note in this doc that states:         

        "Although typically seen for indexes it is also possible that the bug can affect table data."         

        I have used ASSM in several databases, but have postponed some planned tablespace moves to LMT's until I can get the 9.2.0.6 patch applied. I have not actually been bit by this bug, but potential for data corruption was enough to slow me down on implementing ASSM.         

        I haven't seen any performance issues with ASSM, but I'm not sure I've seen any benefits at this point either. I'd like to go back into our data warehouse test server and run some comparisons, but haven't been able to get that effort on the calendar. If I manage to do so, I'll let you know.         

        Robyn                  

        On 6/23/05, John Smiley <jrsmiley_at_gmail.com> wrote:

                I keep hearing claims that ASSM suffers performance problems under heavy DML loads, but I can't find a single solid study that proves this. There are a few isolated bugs documented in MetaLink that discuss problems under specific circumstances, but nothing that speaks to a general performance problem. I found one study that claimed to have proven this, but the proof left a lot to be desired. Other than that, all I've found are vague references to "expert studies." Can anyone point me to some solid research that demonstrates the problem?                  

                Thanks,                                   

		John Smiley
		Technical Management Consultant
		TUSC, Inc.

--
http://www.freelists.org/webpage/oracle-l

Received on Thu Jun 23 2005 - 15:40:12 CDT