The day started off fairly normally – Check GMail for anything that needs immediate attention, then move to blog stats and then hit GrabPERF to see how my sites are behaving. And much to my very pleasant surprise, my blog made it on the front page of GrabPERF and it was on the good (Top 20 performance) side, not the bad side. 🙂 Check out the screenshot below and squint really hard to see Vinny’s blog in there at #17 with an average page load time of 0.4739 seconds.
If you haven’t heard about GrabPERF, it is an awesome free (community supported) service created by Stephen Pierzchala that provides distributed measurement services and monitoring for tracking key performance benchmark of many sites including my blog. The GrabPERF agents gather detailed component, page size, and response code data for the sites they monitor on a regularly scheduled interval and ship it to the central database in real-time where it is available for presentation in the GrabPERF interface. I’ve been a big fan of GrabPERF for a while now and use it as THE key measure for my sites performance.
It was really exciting to see the latest performance results as I’ve had a rough couple of months in terms of hosting. For a while, I had this blog hosted at Kattare to see how their Java/JSP (Tomcat) hosting services stacked up. While my blog was hosted there, I ran into a bug in the awesome Ultimate Tag Warrior 3 WordPress plugin where the plugin filled up my wp_postmeta database table with empty value of meta_value for every post that was viewed and didn’t have one of more tag defined. I had just started using the tag plugin and so not all my posts had tags defined and with the traffic I get, I was causing major issues for Kattare’s shared MySQL database server and so they disabled my site. Since I already had an account with TextDrive and a reliable daily MySQL backup, I moved my site to TextDrive. With the Ultimate Tag Warrior plugin fix, my blog worked for a while but it started causing problems for the folks at TextDrive as I was using the shared hosting feature and the traffic I was getting was adversely affecting other people on my server. So I decided to move to A Small Orange to check out one of their virtual (VPS) offerings to see how they compare to traditional dedicated server. I picked the professional plan which got me 512 MB of RAM, 20 GB of disk space, and 250 GB of bandwidth running CentOS (RedHat Enterprise Linux 4) on a quad 2GHz CPU box for $90 a month. I have been incredibly happy with the performance of the server, the support team and the overall performance of their network and the results from GrabPERF show it.
I am still continuing to ‘play’ with Amazon’s EC2 (Elastic Compute Cloud) offering to see if it could really become the killer solution that will change the hosting landscape. A dedicated (albeit virtual) machine for $70.00 a month is a really compelling story and if Amazon can back that up with additional offerings where you can geographically distribute your applications in multiple datacenters and still scale up/down with computing capacity as needed, why would you host anywhere else? I know Amazon’s EC2 offer is ‘bare-bones’ on purpose where you have to build your server from scratch; you don’t get a web interface like cpanel or plesk to manage your server instance or help with your server configuration once you get up and running. This has to open up opportunities for VAR’s to offer value on top of the EC2 platform by creating ‘hosting-in-a-box’ service where they will build custom Linux deployments, manage them and offer simple management tools. S3, the Amazon storage service has created a huge marketplace for storage, backup tools, online backup vendors and other niche products. I think ECS is going to do the same for the hosting market.
Almost forget: I am running WordPress 2.0.5 with a ton of plugins that are listed on my colophon page and the real difference maker is WP-Cache with this fix to wp-cache-phase2.php that makes it compatible with WordPress 2.0.x.