A big post on Ars Technica’s big redesign. Two or three of my coworkers and I have been working on this for about the last four months, which is pretty exceptional when you think about it.
We basically rebuilt the whole site, 10 years and somehere around 40,000 articles of content, from the ground up.
New servers/configuration, new CMS (Movable Type), and a new design. And we did all this, with two developers, one part-time CSS/HTML guy, and one designer!
This post has a lot of goodies about all the changes we rolled out:
We sat down with ServerCentral and reinvented our server cluster, with a focus on redundancy and speed rather than cutthroat cost efficiency. Located near our Chicago HQ, ServerCentral has very reliable, very robust bandwidth, as we’ve learned over our four years with them. We’re going to be writing about this in more detail in the coming weeks, but we settled on three Dell PE2950s, each with dual 5410 quad core Xeons, 8GB of RAM, and RAID 5 6x73GB storage arrays. Each physical server is running VMWare, which are currently split out into four instances: two lighttpd front ends, a database server—MySQL, and the Movable Type application server (which also hosts Civis, our new login system). We’re currently making plans to double this amount soon as we launch additional dynamic service for our readers. For an OS, we’re using Debian all around, with load balancing courtesy of a pair of Barracudas, and kept tidy by a Fortigate firewall. We continue to rock on with CacheFly, our CDN.
Posted on: 2009 Jan 26
