<?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"
	>
<channel>
	<title>Comments on: EC2, Confluence, S3 and PostgreSQL</title>
	<atom:link href="http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/feed/" rel="self" type="application/rss+xml" />
	<link>http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/</link>
	<description>Notes on things I'm thinking and doing</description>
	<pubDate>Thu, 04 Dec 2008 01:34:13 +0000</pubDate>
	<generator>http://wordpress.org/?v=2.5.1</generator>
		<item>
		<title>By: Ravi</title>
		<link>http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-7973</link>
		<dc:creator>Ravi</dc:creator>
		<pubDate>Fri, 28 Nov 2008 19:35:40 +0000</pubDate>
		<guid isPermaLink="false">http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-7973</guid>
		<description>&lt;p&gt;I am a bit undecided on EBS,&lt;/p&gt;

&lt;p&gt;EBS seems to me like another VM with its instance store exposed as an iSCCI block device, but you pay for the allocation regardless of the usage (0.10 / gb / month).  The reason I say this is that while data may persist- Amazon says that at any point in time it could fail (AFAIK) ... so you would still need to make incremental snapshots out to S3.&lt;/p&gt;

&lt;p&gt;I guess EBS is allows you to pull the logic of keeping data consistent etc. away from machine instances, also EBS blocks cannot be mounted by many instances @ once, I guess the primary instance could export it as nfs.&lt;/p&gt;
</description>
		<content:encoded><![CDATA[<p>I am a bit undecided on EBS,</p>

<p>EBS seems to me like another VM with its instance store exposed as an iSCCI block device, but you pay for the allocation regardless of the usage (0.10 / gb / month).  The reason I say this is that while data may persist- Amazon says that at any point in time it could fail (AFAIK) &#8230; so you would still need to make incremental snapshots out to S3.</p>

<p>I guess EBS is allows you to pull the logic of keeping data consistent etc. away from machine instances, also EBS blocks cannot be mounted by many instances @ once, I guess the primary instance could export it as nfs.</p>]]></content:encoded>
	</item>
	<item>
		<title>By: Tom Davies</title>
		<link>http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-7083</link>
		<dc:creator>Tom Davies</dc:creator>
		<pubDate>Sun, 02 Nov 2008 04:33:17 +0000</pubDate>
		<guid isPermaLink="false">http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-7083</guid>
		<description>&lt;p&gt;@Patrick -- Yes, EBS makes this sort of complex solution redundant.&lt;/p&gt;
</description>
		<content:encoded><![CDATA[<p>@Patrick &#8212; Yes, EBS makes this sort of complex solution redundant.</p>]]></content:encoded>
	</item>
	<item>
		<title>By: Patrick</title>
		<link>http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-7072</link>
		<dc:creator>Patrick</dc:creator>
		<pubDate>Sat, 01 Nov 2008 23:01:59 +0000</pubDate>
		<guid isPermaLink="false">http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-7072</guid>
		<description>&lt;p&gt;A much better idea would be to mount the S3 as a drive on your EC2 instance and point the data directory of postgres to that mount point. Voila! No hassle consistent database.&lt;/p&gt;

&lt;p&gt;This how to shows you how to do it with MySQL - postgerSQL is basically the same method.&lt;/p&gt;

&lt;p&gt;http://www.sunsetlakesoftware.com/2008/09/13/running-drupal-website-amazon-ec2&lt;/p&gt;
</description>
		<content:encoded><![CDATA[<p>A much better idea would be to mount the S3 as a drive on your EC2 instance and point the data directory of postgres to that mount point. Voila! No hassle consistent database.</p>

<p>This how to shows you how to do it with MySQL - postgerSQL is basically the same method.</p>

<p><a href="http://www.sunsetlakesoftware.com/2008/09/13/running-drupal-website-amazon-ec2" rel="nofollow">http://www.sunsetlakesoftware.com/2008/09/13/running-drupal-website-amazon-ec2</a></p>]]></content:encoded>
	</item>
	<item>
		<title>By: Tom Davies</title>
		<link>http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-1559</link>
		<dc:creator>Tom Davies</dc:creator>
		<pubDate>Sat, 12 Apr 2008 11:12:02 +0000</pubDate>
		<guid isPermaLink="false">http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-1559</guid>
		<description>&lt;p&gt;PS: I'm not familiar with MySQL NDBIO...&lt;/p&gt;
</description>
		<content:encoded><![CDATA[<p>PS: I&#8217;m not familiar with MySQL NDBIO&#8230;</p>]]></content:encoded>
	</item>
	<item>
		<title>By: Tom Davies</title>
		<link>http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-1558</link>
		<dc:creator>Tom Davies</dc:creator>
		<pubDate>Sat, 12 Apr 2008 11:11:01 +0000</pubDate>
		<guid isPermaLink="false">http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-1558</guid>
		<description>&lt;p&gt;I never installed a complete setup on an EC2 node, although I did test Postgres backups to S3 from my computer. I'm happy that Postgres backup/restore to S3 is practical.&lt;/p&gt;

&lt;p&gt;I didn't set up a Confluence cluster, but EC2 does allow UDP, so Confluence clustering will work.&lt;/p&gt;

&lt;p&gt;I would recommend that you use the largest machine EC2 provides before resorting to clustering, which introduces some overhead.&lt;/p&gt;

&lt;p&gt;I don't now what the worst case UDP latency between two EC2 nodes is -- you would need to test this (or ask Amazon) before creating a cluster.&lt;/p&gt;
</description>
		<content:encoded><![CDATA[<p>I never installed a complete setup on an EC2 node, although I did test Postgres backups to S3 from my computer. I&#8217;m happy that Postgres backup/restore to S3 is practical.</p>

<p>I didn&#8217;t set up a Confluence cluster, but EC2 does allow UDP, so Confluence clustering will work.</p>

<p>I would recommend that you use the largest machine EC2 provides before resorting to clustering, which introduces some overhead.</p>

<p>I don&#8217;t now what the worst case UDP latency between two EC2 nodes is &#8212; you would need to test this (or ask Amazon) before creating a cluster.</p>]]></content:encoded>
	</item>
	<item>
		<title>By: Matthew Arrott</title>
		<link>http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-1544</link>
		<dc:creator>Matthew Arrott</dc:creator>
		<pubDate>Fri, 11 Apr 2008 21:21:08 +0000</pubDate>
		<guid isPermaLink="false">http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-1544</guid>
		<description>&lt;p&gt;Did you run multiple instance of Conflunence as a fault-tolerant cluster? Did you consider running a MySQL NDBIO cluster for the Fault-Tolerant DB?&lt;/p&gt;

&lt;p&gt;Great to see that some has precede us. Many thanks,&lt;/p&gt;

&lt;p&gt;Matthew&lt;/p&gt;
</description>
		<content:encoded><![CDATA[<p>Did you run multiple instance of Conflunence as a fault-tolerant cluster? Did you consider running a MySQL NDBIO cluster for the Fault-Tolerant DB?</p>

<p>Great to see that some has precede us. Many thanks,</p>

<p>Matthew</p>]]></content:encoded>
	</item>
	<item>
		<title>By: Ram</title>
		<link>http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-1253</link>
		<dc:creator>Ram</dc:creator>
		<pubDate>Tue, 18 Mar 2008 16:47:42 +0000</pubDate>
		<guid isPermaLink="false">http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-1253</guid>
		<description>&lt;p&gt;I would be interested in updates too :)&lt;/p&gt;
</description>
		<content:encoded><![CDATA[<p>I would be interested in updates too <img src='http://diversions.nfshost.com/blog/wp-includes/images/smilies/icon_smile.gif' alt=':)' class='wp-smiley' /> </p>]]></content:encoded>
	</item>
	<item>
		<title>By: Adam Sherman</title>
		<link>http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-778</link>
		<dc:creator>Adam Sherman</dc:creator>
		<pubDate>Thu, 14 Feb 2008 18:54:50 +0000</pubDate>
		<guid isPermaLink="false">http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-778</guid>
		<description>&lt;p&gt;Any news on this project? I'm working on a project now that is PostgreSQL based and needs to be scaled.&lt;/p&gt;

&lt;p&gt;A.&lt;/p&gt;
</description>
		<content:encoded><![CDATA[<p>Any news on this project? I&#8217;m working on a project now that is PostgreSQL based and needs to be scaled.</p>

<p>A.</p>]]></content:encoded>
	</item>
	<item>
		<title>By: Tom Davies</title>
		<link>http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-505</link>
		<dc:creator>Tom Davies</dc:creator>
		<pubDate>Sun, 13 Jan 2008 09:16:30 +0000</pubDate>
		<guid isPermaLink="false">http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-505</guid>
		<description>&lt;p&gt;Thanks for the link, Scott. Interesting that they have 6 months uptime for their DB server.&lt;/p&gt;
</description>
		<content:encoded><![CDATA[<p>Thanks for the link, Scott. Interesting that they have 6 months uptime for their DB server.</p>]]></content:encoded>
	</item>
	<item>
		<title>By: Scott Royston</title>
		<link>http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-484</link>
		<dc:creator>Scott Royston</dc:creator>
		<pubDate>Fri, 11 Jan 2008 02:42:24 +0000</pubDate>
		<guid isPermaLink="false">http://diversions.nfshost.com/blog/2007/12/22/ec2-confluence-s3-and-postgresql/#comment-484</guid>
		<description>&lt;p&gt;I'm very interested in hearing how this goes.  From poking around google it looks like 'GoPlan' is already doing it.
http://comments.deasil.com/2007/05/24/goplan-interview-ec2/&lt;/p&gt;
</description>
		<content:encoded><![CDATA[<p>I&#8217;m very interested in hearing how this goes.  From poking around google it looks like &#8216;GoPlan&#8217; is already doing it.
<a href="http://comments.deasil.com/2007/05/24/goplan-interview-ec2/" rel="nofollow">http://comments.deasil.com/2007/05/24/goplan-interview-ec2/</a></p>]]></content:encoded>
	</item>
</channel>
</rss>
