<?xml version="1.0" encoding="utf-8"?>
<rss xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title>Disqus - Latest Comments for akirkpatrick</title><link>http://disqus.com/by/akirkpatrick/</link><description></description><atom:link href="http://disqus.com/akirkpatrick/comments.rss" rel="self"></atom:link><language>en</language><lastBuildDate>Sun, 25 May 2008 18:58:15 -0000</lastBuildDate><item><title>Re: Twitter At Scale: Will It Work?</title><link>http://www.hurricanesoftwares.com/2008/05/23/twitter-at-scale-will-it-work/#comment-528544</link><description>&lt;p&gt;Twitter mainly uses Ruby for the web front-end. All that back-end critical path stuff is handled by code written in faster languages such as C. This was revealed some weeks ago when there was speculation they'd be dropping RoR.&lt;/p&gt;&lt;p&gt;Google has the luxury of benefiting from a lot of ranking/correlation happening in 'offline' batch runs. This makes the 'online' interactive search code faster to execute. You're right that Twitter's problem is going to need a change in approach for them.&lt;/p&gt;&lt;p&gt;Possibly like what Google does, a way of cementing/optimizing the expected data flows in batch jobs, so when you tweet most of your followers can be reached faster.&lt;/p&gt;&lt;p&gt;As others have speculated, there may be no good way to make an N^2 network scale without opening it up and making it distributed like jabber/email/www.&lt;/p&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">akirkpatrick</dc:creator><pubDate>Sun, 25 May 2008 18:58:15 -0000</pubDate></item></channel></rss>