A few months ago I had some server troubles and switched to Rackspace managed hosting and I can’t praise the reliability and quality of the service enough.
The image above is the uptime graph for wincent.com as measured by Netcraft (click here to see the most up-to-date version of the graph; there have been some reboots for routine maintenance since this article was originally posted but the solidity of the graph is more evident than ever). The diagonal red lines show the steadily growing uptime for the old server which ran FreeBSD. The blue line shows the new server which is hosted by Rackspace and runs Red Hat Enterprise Linux.
In the graph you can see a number of reboots of the FreeBSD system, almost all of them required after updating the operating system with security patches which required a reboot. You’ll also see an impressive uptime of in excess of 400 days at one point. Not bad for a production system on a busy website. I was able to achieve that uptime by conservative system administration (subscribing to security alert mailing lists, applying workarounds immediately, and minimizing changes and unnecessary upgrades to the server). Whenever a security issue popped up which required an upgrade of the kernel or the entire operating system I bit the bullet and performed the upgrade as quickly as possible.
The last two red lines are of interest. These are the only involuntary reboots that you see on the graph. The second to last line corresponds to a server crash in June where the filesystem got corrupted; it was in such an inconsistent state that it wouldn’t reboot when power-cycled. Given the reliability of the software on the system I suspected a faulty hard disk. Unfortunately, the server was on the other side of the world from me, so I had to pay a consultant to fsck the filesystem and I could only hope that it would keep working until I got back to Australia to investigate the problem more thoroughly. It didn’t.
The last red line shows another period which came to an end with another server crash. I suspected that the same dodgy hard disk was the culprit. I just couldn’t stomach the idea of paying $185/hour consultants to get the machine booting again, so I made the jump to Rackspace.
Rackspace is one of the most expensive managed hosting companies around but it is also one of the best; in fact, I would argue that it is the best. Not only do they look after hardware faults like the one I had within an hour, they also take care of restoring things from backup, provide incredible support, and deliver the fastest and most reliable connection to the network that you can find. Rackspace has yet again been ranked the most reliable web host in the latest Netcraft survey (October 2005). Rackspace consistently ranks in the top ten, taking the top slot more often than not.
I don’t know how often I’ll be rebooting that server now that I’m running Red Hat, but one thing you can already see from the graph is the incredible solidity of the connection provided by Rackspace. Notice how the red (FreeBSD) lines break up in some places; I can only presume that this is because my then-ISP in Australia couldn’t deliver the same kind of network reliability that Rackspace does. Check out the blue line; it’s solid as a rock. In most cases you get what you pay for and that’s certainly true of Rackspace. I’m totally happy with their service. If you want the best go with Rackspace. And if you want to support my software development efforts, use this link (I’ll receive a small commission whenever anyone signs up with Rackspace using that link).