Re: 10g RAC using raw devices - curiousity question

From: Martin Klier <usn_at_usn-it.de>
Date: Tue, 25 Sep 2012 08:28:51 +0200
Message-ID: <50614F23.8050402_at_usn-it.de>

Trouble is, that most of those notes and underscore-parameter-manifested events are made for special purposes, to work around bugs and specific issues. You are exposed to special and not regression-tested source code. With every parameter you are growing the rist exponential...

Do as Oracle says: Use them only under supervision of Oracle Support. You might have issues on the other end of the product without knowing why and where.

And in general, IMHO narrowing down the freedom of the CBO ist in 99% of the cases a problem-maker, not a problem-solver.

Regards
Martin

Christopher.Taylor2_at_parallon.net schrieb:
> What is the concern, and what about it concerns you if I were to move it to production?
>
> Chris
>
> -----Original Message-----
> From: Martin Klier [mailto:usn_at_usn-it.de]
> Sent: Monday, September 24, 2012 5:07 PM
> To: Taylor Christopher - Nashville
> Cc: oracle-l_at_freelists.org
> Subject: Re: 10g RAC using raw devices - curiousity question
>
> Uahhh
> =:-O
>
> For a sandbox, it's a brave approach. Hope this isn't production.
>
> Christopher.Taylor2_at_parallon.net schrieb:

>> alter session set workarea_size_policy=manual; alter session set 
>> sort_area_size=1073741824; alter session set 
>> hash_area_size=1073741824; alter session set 
>> "_complex_view_merging"=FALSE;  --- found on Metalink for a different 
>> perf problem & still testing alter session set 
>> "_sort_multiblock_read_count"=256;  --- set for workarea_size_policy 
>> manual alter session set "_hash_multiblock_io_count"=256; --- set for 
>> workarea_size_policy manual alter session set 
>> "_unnest_subquery"=FALSE;  --- found on Metalink for a different perf 
>> problem & still testing alter session set 
>> "_optimizer_use_histograms"=false;  --- Current statistics gathering 
>> strategy needs work alter session set "_optimizer_squ_bottomup"=false;  
>> --- Metalink Note:
>> 1118446.1
>> alter session set "_optimizer_sortmerge_join_enabled"=false; 
>> ---Metalink
>> Note: 444609.1
>> alter session set "_optimizer_join_sel_sanity_check" = true; --- alter 
>> session set "_always_semi_join" = off; alter session set 
>> "_optimizer_max_permutations"=80000; --- alter session set 
>> "_OPTIMIZER_COST_BASED_TRANSFORMATION"=EXHAUSTIVE;
>> ---Metalink Note: 567354.1 -- testing
>> alter session set "_newsort_enabled"=false; --Metalink Bug 6817844.8 
>> -- testing, zero effect observed

>
> --
> Usn's IT Blog for Linux, Oracle, Asterisk http://www.usn-it.de
>
>

-- 
Usn's IT Blog for Linux, Oracle, Asterisk
http://www.usn-it.de

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

Received on Tue Sep 25 2012 - 01:28:51 CDT