<?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: Testing IE8&#8242;s Connection Parallelism</title>
	<atom:link href="http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/feed/" rel="self" type="application/rss+xml" />
	<link>http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/</link>
	<description>A blog by Ryan Breen of CloudFloor</description>
	<lastBuildDate>Thu, 19 May 2011 04:10:34 +0000</lastBuildDate>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
	<generator>http://wordpress.org/?v=3.1</generator>
	<item>
		<title>By: Bruce Stoner</title>
		<link>http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/comment-page-1/#comment-228436</link>
		<dc:creator>Bruce Stoner</dc:creator>
		<pubDate>Mon, 07 Dec 2009 02:34:56 +0000</pubDate>
		<guid isPermaLink="false">http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/#comment-228436</guid>
		<description>Internet Explorer 8 seems to be better than any previous version of IE. IE8 is very stable and rarely crashes or cause blue screens.</description>
		<content:encoded><![CDATA[<p>Internet Explorer 8 seems to be better than any previous version of IE. IE8 is very stable and rarely crashes or cause blue screens.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Lori MacVittie</title>
		<link>http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/comment-page-1/#comment-109393</link>
		<dc:creator>Lori MacVittie</dc:creator>
		<pubDate>Wed, 03 Sep 2008 18:49:58 +0000</pubDate>
		<guid isPermaLink="false">http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/#comment-109393</guid>
		<description>This is an astounding article, thank you!

I&#039;ve &lt;a href=&quot;http://www.slideshare.net/lmacvittie/web-2-and-application-delivery-public/&quot; rel=&quot;nofollow&quot;&gt;spoken&lt;/a&gt; and &lt;a href=&quot;http://devcentral.f5.com/weblogs/macvittie/archive/2008/06/17/3365.aspx&quot; rel=&quot;nofollow&quot;&gt;written&lt;/a&gt; a lot about AJAX and &lt;a href=&quot;http://devcentral.f5.com/weblogs/macvittie/archive/2008/02/21/3086.aspx&quot; rel=&quot;nofollow&quot;&gt;connection management issues&lt;/a&gt; in general and this really puts some hard numbers and empirical data around the subject. 

I haven&#039;t read your blog before, but I&#039;m coming back for more now. 

Great stuff! 
Lori</description>
		<content:encoded><![CDATA[<p>This is an astounding article, thank you!</p>
<p>I&#8217;ve <a href="http://www.slideshare.net/lmacvittie/web-2-and-application-delivery-public/" rel="nofollow">spoken</a> and <a href="http://devcentral.f5.com/weblogs/macvittie/archive/2008/06/17/3365.aspx" rel="nofollow">written</a> a lot about AJAX and <a href="http://devcentral.f5.com/weblogs/macvittie/archive/2008/02/21/3086.aspx" rel="nofollow">connection management issues</a> in general and this really puts some hard numbers and empirical data around the subject. </p>
<p>I haven&#8217;t read your blog before, but I&#8217;m coming back for more now. </p>
<p>Great stuff!<br />
Lori</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Paddy</title>
		<link>http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/comment-page-1/#comment-71004</link>
		<dc:creator>Paddy</dc:creator>
		<pubDate>Fri, 27 Jun 2008 18:31:05 +0000</pubDate>
		<guid isPermaLink="false">http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/#comment-71004</guid>
		<description>All the tests of your connection parallelism apply to users who are far away from the origin server. For example, if your server is in San Jose, whats the effect of 4 parallel connections for San Jose users (guess? its gonna be worse than  single connection)

Look at their bandwidth-delay product and knowing full well that TCP cannot send less than 1MSS, 4 connections will have to tranmit the slowest at 4*1KB *8b/B/RTT = 32 Kb/RTT and RTT here is very small like 10ms. This means queuing will inevitable happen.

try the opening page of www.hotmail.com from san jose to see what I mean..

That said, you results are very relevant for dissemination of content from a single server. For distributed servers, people close enough to the edges, this would be counter-productive.</description>
		<content:encoded><![CDATA[<p>All the tests of your connection parallelism apply to users who are far away from the origin server. For example, if your server is in San Jose, whats the effect of 4 parallel connections for San Jose users (guess? its gonna be worse than  single connection)</p>
<p>Look at their bandwidth-delay product and knowing full well that TCP cannot send less than 1MSS, 4 connections will have to tranmit the slowest at 4*1KB *8b/B/RTT = 32 Kb/RTT and RTT here is very small like 10ms. This means queuing will inevitable happen.</p>
<p>try the opening page of <a href="http://www.hotmail.com" rel="nofollow">http://www.hotmail.com</a> from san jose to see what I mean..</p>
<p>That said, you results are very relevant for dissemination of content from a single server. For distributed servers, people close enough to the edges, this would be counter-productive.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Ryan Breen</title>
		<link>http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/comment-page-1/#comment-60889</link>
		<dc:creator>Ryan Breen</dc:creator>
		<pubDate>Mon, 09 Jun 2008 12:15:31 +0000</pubDate>
		<guid isPermaLink="false">http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/#comment-60889</guid>
		<description>Mark,

Being wrong made me think there was a limit to concurrent connections.  It&#039;s a longstanding misconception I had, but it has since been corrected.  The only limit is per host.

I don&#039;t see any issues with my test pages.  I have intermittent problems with dreamhost, but everything appears to be functional now.</description>
		<content:encoded><![CDATA[<p>Mark,</p>
<p>Being wrong made me think there was a limit to concurrent connections.  It&#8217;s a longstanding misconception I had, but it has since been corrected.  The only limit is per host.</p>
<p>I don&#8217;t see any issues with my test pages.  I have intermittent problems with dreamhost, but everything appears to be functional now.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Mark</title>
		<link>http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/comment-page-1/#comment-60774</link>
		<dc:creator>Mark</dc:creator>
		<pubDate>Mon, 09 Jun 2008 07:52:10 +0000</pubDate>
		<guid isPermaLink="false">http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/#comment-60774</guid>
		<description>What makes you think WinINET has any limit to concurrent connections at all?  

your test pages are broken, btw.</description>
		<content:encoded><![CDATA[<p>What makes you think WinINET has any limit to concurrent connections at all?  </p>
<p>your test pages are broken, btw.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Biondi Tang</title>
		<link>http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/comment-page-1/#comment-51336</link>
		<dc:creator>Biondi Tang</dc:creator>
		<pubDate>Wed, 21 May 2008 06:27:44 +0000</pubDate>
		<guid isPermaLink="false">http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/#comment-51336</guid>
		<description>I used IE8 getting the same page at http://demo.ajaxperformance.com and saw a quite different graph in Page Detailer.  The number of parallel connections started from 3 in the first round, 5 the second, 9 the third, and another 9 the fourth, for the total of 26 images.

While running in IE7 Emulate mode, a different graph with 6, 6, 4, 4, 6 parallel connections in 5 rounds.

I think there might be some other factors affecting the result such as file size, internet connection bandwidth, local cache...etc.

I was considering to use &quot;web accelerator&quot; appliance that do the same job by layer 7 protocol.  It seems that this &quot;multi-connect&quot; acceleration will be replaced by next version of browsers.  Any similar case to share?</description>
		<content:encoded><![CDATA[<p>I used IE8 getting the same page at <a href="http://demo.ajaxperformance.com" rel="nofollow">http://demo.ajaxperformance.com</a> and saw a quite different graph in Page Detailer.  The number of parallel connections started from 3 in the first round, 5 the second, 9 the third, and another 9 the fourth, for the total of 26 images.</p>
<p>While running in IE7 Emulate mode, a different graph with 6, 6, 4, 4, 6 parallel connections in 5 rounds.</p>
<p>I think there might be some other factors affecting the result such as file size, internet connection bandwidth, local cache&#8230;etc.</p>
<p>I was considering to use &#8220;web accelerator&#8221; appliance that do the same job by layer 7 protocol.  It seems that this &#8220;multi-connect&#8221; acceleration will be replaced by next version of browsers.  Any similar case to share?</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Steve Laniel</title>
		<link>http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/comment-page-1/#comment-39731</link>
		<dc:creator>Steve Laniel</dc:creator>
		<pubDate>Mon, 17 Mar 2008 12:51:20 +0000</pubDate>
		<guid isPermaLink="false">http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/#comment-39731</guid>
		<description>I&#039;m surprised the browser can&#039;t learn what the optimum number of parallel connections is, and adjust over time. That seems like an obvious next step. I wonder if other browsers will learn from this performance analysis and do IE one better.</description>
		<content:encoded><![CDATA[<p>I&#8217;m surprised the browser can&#8217;t learn what the optimum number of parallel connections is, and adjust over time. That seems like an obvious next step. I wonder if other browsers will learn from this performance analysis and do IE one better.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: andrew</title>
		<link>http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/comment-page-1/#comment-39724</link>
		<dc:creator>andrew</dc:creator>
		<pubDate>Mon, 17 Mar 2008 11:57:59 +0000</pubDate>
		<guid isPermaLink="false">http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/#comment-39724</guid>
		<description>I use Opera browser. I can set up up to 128 (!) connection per server, and up to 128 connections total.

Just choose Tools-Preferences-Advanced-Network.

The defaults are 8 and 20 respectivly, and I don&#039;t know why.</description>
		<content:encoded><![CDATA[<p>I use Opera browser. I can set up up to 128 (!) connection per server, and up to 128 connections total.</p>
<p>Just choose Tools-Preferences-Advanced-Network.</p>
<p>The defaults are 8 and 20 respectivly, and I don&#8217;t know why.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Pelican</title>
		<link>http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/comment-page-1/#comment-39678</link>
		<dc:creator>Pelican</dc:creator>
		<pubDate>Mon, 17 Mar 2008 02:30:22 +0000</pubDate>
		<guid isPermaLink="false">http://www.ajaxperformance.com/2008/03/16/testing-ie8s-connection-parallelism/#comment-39678</guid>
		<description>Lovely article. I keep following your connection stuff and cannot believe it is 18 now..I use Gomez a lot and those numbers definitely make sense..BTW do you have any official confirmation of this from the IE team?</description>
		<content:encoded><![CDATA[<p>Lovely article. I keep following your connection stuff and cannot believe it is 18 now..I use Gomez a lot and those numbers definitely make sense..BTW do you have any official confirmation of this from the IE team?</p>
]]></content:encoded>
	</item>
</channel>
</rss>

<!-- Dynamic Page Served (once) in 0.197 seconds -->

