<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>IO Storage &amp; Data Path on URE</title><link>https://ure.us/pillars/io-storage--data-path/</link><description>Recent content in IO Storage &amp; Data Path on URE</description><generator>Hugo -- 0.164.0</generator><language>en-us</language><lastBuildDate>Thu, 02 Apr 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://ure.us/pillars/io-storage--data-path/index.xml" rel="self" type="application/rss+xml"/><item><title>A Million PDFs Won't Organize Themselves</title><link>https://ure.us/articles/a-million-pdfs-wont-organize-themselves/</link><pubDate>Thu, 02 Apr 2026 00:00:00 +0000</pubDate><guid>https://ure.us/articles/a-million-pdfs-wont-organize-themselves/</guid><description>A million corporate PDFs in a RAG pipeline revealed what tutorials never mention -- the parsing, embedding, and graph labor that no shortcut survives at scale.</description></item><item><title>Tail Latency Killed My Beowulf Cluster in 2006</title><link>https://ure.us/articles/tail-latency-killed-beowulf-cluster-2006/</link><pubDate>Sun, 04 Jan 2026 00:00:00 +0000</pubDate><guid>https://ure.us/articles/tail-latency-killed-beowulf-cluster-2006/</guid><description>In 2006, I learned that scaling out doesn&amp;#39;t work when the interconnect is the bottleneck. Twenty years later, the same physics governs GPU infrastructure.</description></item></channel></rss>