<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
		>
<channel>
	<title>Comments on: SQL Server 2000 and AWE Memory</title>
	<atom:link href="http://www.jimmcleod.net/blog/index.php/2008/06/09/sql-server-2000-and-awe-memory/feed/" rel="self" type="application/rss+xml" />
	<link>http://www.jimmcleod.net/blog/index.php/2008/06/09/sql-server-2000-and-awe-memory/</link>
	<description>Musings about SQL, databases, and my world in general...</description>
	<lastBuildDate>Tue, 20 Dec 2011 15:37:13 +1000</lastBuildDate>
	<generator>http://wordpress.org/?v=2.8.6</generator>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
		<item>
		<title>By: MikeJ</title>
		<link>http://www.jimmcleod.net/blog/index.php/2008/06/09/sql-server-2000-and-awe-memory/comment-page-1/#comment-10121</link>
		<dc:creator>MikeJ</dc:creator>
		<pubDate>Fri, 24 Apr 2009 11:35:36 +0000</pubDate>
		<guid isPermaLink="false">http://www.jimmcleod.net/blog/index.php/2008/06/09/sql-server-2000-and-awe-memory/#comment-10121</guid>
		<description>Hi,

Not exactly sure how enabling AWE is assisting you. Usually the memory above 4 Gb is entirely allocated as Buffer Cache. This can cover up I/O problems on the Server as the page read rate drops as more of the pages are in the buffer cache.

Poor performance is usually caused by
1. SQL running that does a tablescan rather than using an index - check the execution plan s for the most commonly running SQL that an index is being used to access the table.
2, Statistics are not up to date or are inaccurate - consider switching autoupdate of statistics off and running a maintenance job that does it for each table in the application. If the tables are very big then consider an update of statitics with fullscan.
3. Are the Indexes badly fragmented or don&#039;t match the configured fill factors - Run a dbcc dbreindex on the affected tables
4. Do you have excessive OS fragmentation on the tablespaces? Consider running a defrag of the disk
5. You could also experiment with the Degree of Parallelism and see if it improves the execution times - This can help if the DB Server CPU goes through the roof

Regards

Mike</description>
		<content:encoded><![CDATA[<p>Hi,</p>
<p>Not exactly sure how enabling AWE is assisting you. Usually the memory above 4 Gb is entirely allocated as Buffer Cache. This can cover up I/O problems on the Server as the page read rate drops as more of the pages are in the buffer cache.</p>
<p>Poor performance is usually caused by<br />
1. SQL running that does a tablescan rather than using an index &#8211; check the execution plan s for the most commonly running SQL that an index is being used to access the table.<br />
2, Statistics are not up to date or are inaccurate &#8211; consider switching autoupdate of statistics off and running a maintenance job that does it for each table in the application. If the tables are very big then consider an update of statitics with fullscan.<br />
3. Are the Indexes badly fragmented or don&#8217;t match the configured fill factors &#8211; Run a dbcc dbreindex on the affected tables<br />
4. Do you have excessive OS fragmentation on the tablespaces? Consider running a defrag of the disk<br />
5. You could also experiment with the Degree of Parallelism and see if it improves the execution times &#8211; This can help if the DB Server CPU goes through the roof</p>
<p>Regards</p>
<p>Mike</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: MA Roberts</title>
		<link>http://www.jimmcleod.net/blog/index.php/2008/06/09/sql-server-2000-and-awe-memory/comment-page-1/#comment-1401</link>
		<dc:creator>MA Roberts</dc:creator>
		<pubDate>Fri, 18 Jul 2008 14:56:31 +0000</pubDate>
		<guid isPermaLink="false">http://www.jimmcleod.net/blog/index.php/2008/06/09/sql-server-2000-and-awe-memory/#comment-1401</guid>
		<description>So...we&#039;ve been having some performance issues with our SQL Server 2005 box - it&#039;s running an SAP ERP solution and certain processes take a long, LONG time (upwards of five minutes) to execute. The box is a Windows Server 2003 Standard Edition machine with 4 GB of RAM (we only have about 30 employees, only half of which are ever using the ERP system at any time). We&#039;ve been running performance logs and whatnot to see which queries are causing the problems...the big problem is a bunch of queries running against a table with 350,000 lines - the queries are not optimized correctly and as far as I&#039;m able to determine the execution plan is underestimating the number of lines it needs to pull...creating a bunch of page faults, excessive disk reads, etc.

So our IT guy (who&#039;s not a SQL expert) noticed that turning on AWE created a temporary increase in performance when some of these slowdowns occur. Which - OK, we have only 4GB of RAM, so it&#039;s obviously not enabling access to more than 4GB, like AWE is designed to do. Since AWE turns off some of SQL Server&#039;s dynamic memory addressing, I&#039;m sure that at some point it&#039;s actually hurting our performance - because the performance problems return later, even with AWE switched on.

I figure that the AWE switch must be doing something ELSE when it&#039;s switched off / on to the way the system operates - clearing the buffer cache, or something, to clear up some of the bottlenecks. My concern would be that it&#039;s clearing the ENTIRE buffer cache, including uncommitted transactions or data that a checkpoint hasn&#039;t written to disk yet...regardless, it&#039;s clear that he doesn&#039;t have a firm grasp on what the AWE switch is doing to reconfigure memory when it&#039;s turned on in the middle of business operations and users being logged in, etc. (something that, at the very least, should be tested before being rolled out).

So - what does the AWE switch DO, exactly, to the buffer cache and other running processes when it&#039;s turned on in medias res? I can&#039;t find any information about this anywhere.</description>
		<content:encoded><![CDATA[<p>So&#8230;we&#8217;ve been having some performance issues with our SQL Server 2005 box &#8211; it&#8217;s running an SAP ERP solution and certain processes take a long, LONG time (upwards of five minutes) to execute. The box is a Windows Server 2003 Standard Edition machine with 4 GB of RAM (we only have about 30 employees, only half of which are ever using the ERP system at any time). We&#8217;ve been running performance logs and whatnot to see which queries are causing the problems&#8230;the big problem is a bunch of queries running against a table with 350,000 lines &#8211; the queries are not optimized correctly and as far as I&#8217;m able to determine the execution plan is underestimating the number of lines it needs to pull&#8230;creating a bunch of page faults, excessive disk reads, etc.</p>
<p>So our IT guy (who&#8217;s not a SQL expert) noticed that turning on AWE created a temporary increase in performance when some of these slowdowns occur. Which &#8211; OK, we have only 4GB of RAM, so it&#8217;s obviously not enabling access to more than 4GB, like AWE is designed to do. Since AWE turns off some of SQL Server&#8217;s dynamic memory addressing, I&#8217;m sure that at some point it&#8217;s actually hurting our performance &#8211; because the performance problems return later, even with AWE switched on.</p>
<p>I figure that the AWE switch must be doing something ELSE when it&#8217;s switched off / on to the way the system operates &#8211; clearing the buffer cache, or something, to clear up some of the bottlenecks. My concern would be that it&#8217;s clearing the ENTIRE buffer cache, including uncommitted transactions or data that a checkpoint hasn&#8217;t written to disk yet&#8230;regardless, it&#8217;s clear that he doesn&#8217;t have a firm grasp on what the AWE switch is doing to reconfigure memory when it&#8217;s turned on in the middle of business operations and users being logged in, etc. (something that, at the very least, should be tested before being rolled out).</p>
<p>So &#8211; what does the AWE switch DO, exactly, to the buffer cache and other running processes when it&#8217;s turned on in medias res? I can&#8217;t find any information about this anywhere.</p>
]]></content:encoded>
	</item>
</channel>
</rss>

