<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Zram on Tony Metzidis</title>
    <link>https://tonym.us/tag/zram.html</link>
    <description>Recent content in Zram on Tony Metzidis</description>
    <generator>Hugo</generator>
    <language>en-us</language>
    <lastBuildDate>Thu, 01 Oct 2026 22:30:00 -0700</lastBuildDate>
    <atom:link href="https://tonym.us/tag/zram/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Surviving the 1 GB Squeeze: Engineering an Unstoppable Alpine GCE Image</title>
      <link>https://tonym.us/alpine-gce-extreme-memory-survival.html</link>
      <pubDate>Thu, 01 Oct 2026 22:30:00 -0700</pubDate>
      <guid>https://tonym.us/alpine-gce-extreme-memory-survival.html</guid>
      <description>&lt;p&gt;In my previous post, we looked at how default Google guest daemons consume over 15% of available memory on a 1 GB Google Compute Engine (&lt;code&gt;e2-micro&lt;/code&gt;) instance, pushing standard Debian images into SSH lockout during moderate memory pressure.&lt;/p&gt;&#xA;&lt;p&gt;To see how far we could push the lower boundary of GCE instances without sacrificing reliability, I built a custom minimal Alpine Linux image (&lt;code&gt;alpine-3-24-1-uefi-lite-v5&lt;/code&gt;).&lt;/p&gt;&#xA;&lt;p&gt;We subjected this image to a punishing synthetic stress harness: pinning &lt;strong&gt;900 MB to 1,250 MB of memory&lt;/strong&gt;, driving &lt;strong&gt;16 concurrent worker threads&lt;/strong&gt;, and executing &lt;strong&gt;over 167,000 sub-processes&lt;/strong&gt; in five minutes. While standard Debian froze solid, the Alpine instance maintained sub-second SSH logins and responsive interactive keystrokes.&lt;/p&gt;</description>
    </item>
  </channel>
</rss>
