<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Configuration on John Hamelink</title>
    <link>https://johnhame.link/tags/configuration/</link>
    <description>Recent content in Configuration on John Hamelink</description>
    <generator>Hugo</generator>
    <language>en</language>
    <copyright>John hamelink</copyright>
    <lastBuildDate>Mon, 21 Sep 2015 13:18:00 +0100</lastBuildDate>
    <atom:link href="https://johnhame.link/tags/configuration/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Pull-based Deployment at Find a Player</title>
      <link>https://johnhame.link/posts/pull-based-deployment-at-find-a-player/</link>
      <pubDate>Mon, 21 Sep 2015 13:18:00 +0100</pubDate>
      <guid>https://johnhame.link/posts/pull-based-deployment-at-find-a-player/</guid>
      <description>&lt;p&gt;Most startups care a lot about scalability. &lt;a href=&#34;http://findaplayer.com&#34;&gt;Find a player&lt;/a&gt; is no different. A lot&#xA;of people have asked me how I went about designing Find a Player&amp;rsquo;s deployment&#xA;system, so I thought I&amp;rsquo;d document it here at a high level.&lt;/p&gt;&#xA;&lt;h2 id=&#34;background&#34;&gt;Background&lt;/h2&gt;&#xA;&lt;p&gt;Before we embarked on this project, we were using ansible to handle push-based&#xA;deployments to multiple servers. However, when we looked into using an&#xA;autoscaler in AWS to handle traffic spikes, we realised that a push-based&#xA;deployment no longer makes sense as it requires an external widget to run the&#xA;deploy on a server if it is spawned. It became very clear that it made much more&#xA;sense to have the spawned server itself run the deploy locally on itself.&lt;/p&gt;</description>
    </item>
  </channel>
</rss>
