<?xml version="1.0" encoding="UTF-8" ?>
<?xml-stylesheet type="text/xsl" href="http://blogs.msdn.com/utility/FeedStylesheets/rss.xsl" media="screen"?><rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:slash="http://purl.org/rss/1.0/modules/slash/" xmlns:wfw="http://wellformedweb.org/CommentAPI/"><channel><title>Modeling servers for resource consumption</title><link>http://blogs.msdn.com/patrick_dussud/archive/2006/12/05/modeling-servers-for-resource-consumption.aspx</link><description>In the past few years, I have helped various groups deal with server problems related to resource usage. A typical scenario goes like this “we have a server we are about to release. It does great on our internal stress tests – generally composed of a</description><dc:language>en</dc:language><generator>CommunityServer 2.1 SP1 (Build: 61025.2)</generator><item><title>re: Modeling servers for resource consumption</title><link>http://blogs.msdn.com/patrick_dussud/archive/2006/12/05/modeling-servers-for-resource-consumption.aspx#1242621</link><pubDate>Sat, 09 Dec 2006 02:57:07 GMT</pubDate><guid isPermaLink="false">91d46819-8472-40ad-a661-2c78acb4018c:1242621</guid><dc:creator>pradeepchellappan</dc:creator><description>&lt;p&gt;Could you please change the font, its almost unreadable :(&lt;/p&gt;
</description></item><item><title>New and Notable 132</title><link>http://blogs.msdn.com/patrick_dussud/archive/2006/12/05/modeling-servers-for-resource-consumption.aspx#1358153</link><pubDate>Sun, 24 Dec 2006 19:47:04 GMT</pubDate><guid isPermaLink="false">91d46819-8472-40ad-a661-2c78acb4018c:1358153</guid><dc:creator>Sam Gentile</dc:creator><description>&lt;p&gt;A special Christmas Eve version to clean out my Feed Demon box as I will be on vacation next week and&lt;/p&gt;
</description></item><item><title>re: Modeling servers for resource consumption</title><link>http://blogs.msdn.com/patrick_dussud/archive/2006/12/05/modeling-servers-for-resource-consumption.aspx#1409313</link><pubDate>Thu, 04 Jan 2007 07:54:50 GMT</pubDate><guid isPermaLink="false">91d46819-8472-40ad-a661-2c78acb4018c:1409313</guid><dc:creator>sambo99</dc:creator><description>&lt;p&gt;Second vote on the font, white on black is really hard to read. &lt;/p&gt;
&lt;p&gt;We have also seen many issues with resource usage in the past with our app. DataSets are a really common way to get these issues, you fill up a DataSet with &amp;quot;select * from table&amp;quot; which in Dev is really small but in production has millions of rows, and boom. &lt;/p&gt;
&lt;p&gt;These issues are really tricky to debug, in general we use windbg + sos to debug this stuff. Scitech's &amp;quot;.net memory profiler&amp;quot; is another option, but in general windbg is faster. &lt;/p&gt;
&lt;p&gt;In some rare instances we have seen heap fragmentation, but this is not the general case. &lt;/p&gt;
&lt;p&gt;The hardest issues we have had to debug are the unmanaged leaks in our app. &lt;/p&gt;
&lt;p&gt;Regarding slow communication lines, many of the more expensive hardware routers can be configured to simulate a slow link. There are also a few software options out there. Perhaps the underlying libraries should also provide the ability to run in test mode where they slow down comms for all the .net provided comms stuff. &lt;/p&gt;
&lt;p&gt;An idea that I think is probably not that good but may be worth exploring would be for the GC to swap Gen2 objects to the disk if its nearing fatal mem usage (swapping least used objects first) &lt;/p&gt;
</description></item><item><title>re: Modeling servers for resource consumption</title><link>http://blogs.msdn.com/patrick_dussud/archive/2006/12/05/modeling-servers-for-resource-consumption.aspx#1623270</link><pubDate>Thu, 08 Feb 2007 05:17:43 GMT</pubDate><guid isPermaLink="false">91d46819-8472-40ad-a661-2c78acb4018c:1623270</guid><dc:creator>Patrick Dussud</dc:creator><description>&lt;p&gt;This is a very good comment. I am preparing a blog entry to give techniques to deal with the scenario you are talking about, within certain limits. Persisting objects to the disk under the cover is hard. We don't know which objects are used the least and it is hard to be totally transparent. Obviously this is possible, a number of in memory object databases are doing just this but it always entail a non trivial performance loss for the common case. Some application specific measures are generally cheaper and can be targeted to just a specific part of the object population where the tradeoff between performance and memory footprint can be shifted correctly &amp;nbsp;&lt;/p&gt;
</description></item><item><title>Limiting resource consumption on servers. </title><link>http://blogs.msdn.com/patrick_dussud/archive/2006/12/05/modeling-servers-for-resource-consumption.aspx#2490753</link><pubDate>Wed, 09 May 2007 03:19:49 GMT</pubDate><guid isPermaLink="false">91d46819-8472-40ad-a661-2c78acb4018c:2490753</guid><dc:creator>CLR, architectures, and stuff</dc:creator><description>&lt;p&gt;In my last blog entry I described the problem of servers over committing resources. There are several&lt;/p&gt;
</description></item><item><title>re: Modeling servers for resource consumption</title><link>http://blogs.msdn.com/patrick_dussud/archive/2006/12/05/modeling-servers-for-resource-consumption.aspx#8618416</link><pubDate>Thu, 19 Jun 2008 00:23:25 GMT</pubDate><guid isPermaLink="false">91d46819-8472-40ad-a661-2c78acb4018c:8618416</guid><dc:creator>dipbi</dc:creator><description>&lt;p&gt;Could you comment on garbage collection mechanisms on network performance monitoring agents and their effect on management of agent lifetime and application throughput ?&lt;/p&gt;
</description></item><item><title>New and Notable 132</title><link>http://blogs.msdn.com/patrick_dussud/archive/2006/12/05/modeling-servers-for-resource-consumption.aspx#9165677</link><pubDate>Tue, 02 Dec 2008 21:59:59 GMT</pubDate><guid isPermaLink="false">91d46819-8472-40ad-a661-2c78acb4018c:9165677</guid><dc:creator>Sam Gentile's Blog</dc:creator><description>&lt;p&gt;A special Christmas Eve version to clean out my Feed Demon box as I will be on vacation next week and then in Belgium for a week after that so don&amp;amp;#39;t expect much if any blogging. CLR Modeling servers for resource consumption Windows keyed events, critical&lt;/p&gt;
</description></item></channel></rss>