<?xml version="1.0" encoding="UTF-8"?><!-- generator="wordpress/2.3" -->
<rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	>
<channel>
	<title>Comments on: Performance - Transfering Stats 11i</title>
	<link>http://onlineappsdba.com/index.php/2008/01/22/performance-transfering-stats-11i/</link>
	<description>One Stop Shop for Apps DBA's</description>
	<pubDate>Thu, 09 Sep 2010 21:46:42 +0000</pubDate>
	<generator>http://wordpress.org/?v=2.3</generator>
		<item>
		<title>By: Anthony Cordell</title>
		<link>http://onlineappsdba.com/index.php/2008/01/22/performance-transfering-stats-11i/#comment-2905</link>
		<dc:creator>Anthony Cordell</dc:creator>
		<pubDate>Thu, 24 Jan 2008 22:56:22 +0000</pubDate>
		<guid>http://onlineappsdba.com/index.php/2008/01/22/performance-transfering-stats-11i/#comment-2905</guid>
		<description>Using the copy command in SQL*Plus translates all number values to decimals thus losing precision.  At a minimum, this will cause the clustering factors to be either 1 or 0 instead of the in-between values that they probably are on the source.  I'd recommend using exp/imp or a dblink for the actual transfer.

Also, you would want to include your CPU costing information to make sure plans are identical across environments.  export_system_stats/import_system_stats can be used the same way.</description>
		<content:encoded><![CDATA[<p>Using the copy command in SQL*Plus translates all number values to decimals thus losing precision.  At a minimum, this will cause the clustering factors to be either 1 or 0 instead of the in-between values that they probably are on the source.  I&#8217;d recommend using exp/imp or a dblink for the actual transfer.</p>
<p>Also, you would want to include your CPU costing information to make sure plans are identical across environments.  export_system_stats/import_system_stats can be used the same way.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Ramnik Gupta</title>
		<link>http://onlineappsdba.com/index.php/2008/01/22/performance-transfering-stats-11i/#comment-2872</link>
		<dc:creator>Ramnik Gupta</dc:creator>
		<pubDate>Wed, 23 Jan 2008 18:41:17 +0000</pubDate>
		<guid>http://onlineappsdba.com/index.php/2008/01/22/performance-transfering-stats-11i/#comment-2872</guid>
		<description>Please find my comments below:

In 8i with Apps Environment - Never gather stats for the SYS schema, because  it is not supported or recommended

In 9i with Apps Environment - Gathering status for SYS schema is supported but not necessary.

In 10g with Apps Environment - It is mandatory to collect stats for SYS schema

Metalink note # 368252.1 is the reason to believe. I hope this will close the loop.</description>
		<content:encoded><![CDATA[<p>Please find my comments below:</p>
<p>In 8i with Apps Environment - Never gather stats for the SYS schema, because  it is not supported or recommended</p>
<p>In 9i with Apps Environment - Gathering status for SYS schema is supported but not necessary.</p>
<p>In 10g with Apps Environment - It is mandatory to collect stats for SYS schema</p>
<p>Metalink note # 368252.1 is the reason to believe. I hope this will close the loop.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Jagjeet Singh</title>
		<link>http://onlineappsdba.com/index.php/2008/01/22/performance-transfering-stats-11i/#comment-2870</link>
		<dc:creator>Jagjeet Singh</dc:creator>
		<pubDate>Wed, 23 Jan 2008 17:23:53 +0000</pubDate>
		<guid>http://onlineappsdba.com/index.php/2008/01/22/performance-transfering-stats-11i/#comment-2870</guid>
		<description>Yes Ramnik,

I have read the post and I am saying there would be some scenarios where system would 
run much faster if you analyze the SYS schema.

Performance impact can be noticed on a system with
really large no. of records in internal tables.
Mean with really large no. of objects.

If CBO is useful for the Non-SYS schemas then
this would also be helpful for sys schema
as internal/recursive queries won't perform
good with hard-coded hints [ RBO ].

This is the reason oracle has introduced this
feature.

1 - May I know what made you to believe on this, If you can tell us the reason ..


&#62;It is possible to transfer statistics between &#62;servers allowing consistent execution plans &#62;between servers

Apart from statistics there are other environment settings also which may change the
plan of query.

Regards,
Jagjeet Singh</description>
		<content:encoded><![CDATA[<p>Yes Ramnik,</p>
<p>I have read the post and I am saying there would be some scenarios where system would<br />
run much faster if you analyze the SYS schema.</p>
<p>Performance impact can be noticed on a system with<br />
really large no. of records in internal tables.<br />
Mean with really large no. of objects.</p>
<p>If CBO is useful for the Non-SYS schemas then<br />
this would also be helpful for sys schema<br />
as internal/recursive queries won&#8217;t perform<br />
good with hard-coded hints [ RBO ].</p>
<p>This is the reason oracle has introduced this<br />
feature.</p>
<p>1 - May I know what made you to believe on this, If you can tell us the reason ..</p>
<p>&gt;It is possible to transfer statistics between &gt;servers allowing consistent execution plans &gt;between servers</p>
<p>Apart from statistics there are other environment settings also which may change the<br />
plan of query.</p>
<p>Regards,<br />
Jagjeet Singh</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Ramnik Gupta</title>
		<link>http://onlineappsdba.com/index.php/2008/01/22/performance-transfering-stats-11i/#comment-2867</link>
		<dc:creator>Ramnik Gupta</dc:creator>
		<pubDate>Wed, 23 Jan 2008 15:05:47 +0000</pubDate>
		<guid>http://onlineappsdba.com/index.php/2008/01/22/performance-transfering-stats-11i/#comment-2867</guid>
		<description>Hi Jagjeet Singh,

Please read the post care fully.

"It is found gathering stats for the SYS schema "can" make the system run slower, not faster."

I clearly mentioned that it can make the system run slower not faster. 

We have 7 TB of database size and this post is based on my experience and Oracle Recommdation. It doesn't apply to all environment (small or medium). It is really good you pointed out. Do you still have question please ask.

Regards
Ramnik Gupta</description>
		<content:encoded><![CDATA[<p>Hi Jagjeet Singh,</p>
<p>Please read the post care fully.</p>
<p>&#8220;It is found gathering stats for the SYS schema &#8220;can&#8221; make the system run slower, not faster.&#8221;</p>
<p>I clearly mentioned that it can make the system run slower not faster. </p>
<p>We have 7 TB of database size and this post is based on my experience and Oracle Recommdation. It doesn&#8217;t apply to all environment (small or medium). It is really good you pointed out. Do you still have question please ask.</p>
<p>Regards<br />
Ramnik Gupta</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Jagjeet Singh</title>
		<link>http://onlineappsdba.com/index.php/2008/01/22/performance-transfering-stats-11i/#comment-2863</link>
		<dc:creator>Jagjeet Singh</dc:creator>
		<pubDate>Wed, 23 Jan 2008 11:20:08 +0000</pubDate>
		<guid>http://onlineappsdba.com/index.php/2008/01/22/performance-transfering-stats-11i/#comment-2863</guid>
		<description>&#62;It is found gathering stats for the SYS schema can make the system run slower, not faster.

It is not the case always that gather sys stats
will make system slower. Please share your exp.
on this with specific detail. you can not see much
effect for medium size databases but databases
with large no. of objects would get benefited. 

In 9i -- Gather stats on sys only be done after
        consulting with oracle.

in 10g -- we can gather stats on sys as oracle
          says it would use CBO for all recursive queries BUT I have seen in some queries still hard-coded hints.

11g    -- I hope this is purely CBO for sys/internal queries.

Regards,
Jagjeet Singh</description>
		<content:encoded><![CDATA[<p>&gt;It is found gathering stats for the SYS schema can make the system run slower, not faster.</p>
<p>It is not the case always that gather sys stats<br />
will make system slower. Please share your exp.<br />
on this with specific detail. you can not see much<br />
effect for medium size databases but databases<br />
with large no. of objects would get benefited. </p>
<p>In 9i &#8212; Gather stats on sys only be done after<br />
        consulting with oracle.</p>
<p>in 10g &#8212; we can gather stats on sys as oracle<br />
          says it would use CBO for all recursive queries BUT I have seen in some queries still hard-coded hints.</p>
<p>11g    &#8212; I hope this is purely CBO for sys/internal queries.</p>
<p>Regards,<br />
Jagjeet Singh</p>
]]></content:encoded>
	</item>
</channel>
</rss>

<!-- Dynamic Page Served (once) in 0.168 seconds -->
