A new survey from the Yankee Group reveals some telling information about the reliability of Linux, Unix, and Windows servers. Hint: Redmond isn’t going to be happy.
If you were to be asked, out of all the server operating systems on the market, which platform had the highest reliability (in terms of downtime), what would your guess be?
We’re talking all of the heavy hitters, mind you: Linux, Unix and even some Windows versions thrown in. Still not sure? According to the Yankee Group’s second annual 2007-2008 Server OS Reliability Survey, the hands-down winner for reliability is … IBM’s AIX.
Conducted by the Yankee Group and first reported by the Institute for Advanced Professional Studies, the independent survey examined 700 users in 27 countries to measure the hourly per year downtime of all of the major server operating systems. The summary of the survey reveals some telling information about the state of reliability in the server space today.
First off, as a class Unix servers performed consistently better than other operating systems in the survey. AIX boxes came in at just 36 minutes of yearly downtime, as measured in 2007. HP-UX came in with 1.18 hours of downtime, and Solaris clocked 1.44 hours.
This is not to say the Linux distributions were shoddy in any way. Red Hat Linux installs (with customizations) had about 52 minutes of downtime, while out-of-the-box installations of Red Hat reported 1.73 hours. SUSE Linux-based servers were surveyed at 1.08 of downtime per year.
Curiously, the category of “Other Linux” (Turbolinux and Mandriva included) appeared to have stats even lower than AIX’s, with just 23 minutes of downtime. I am not sure why, with this number, the Other Linux category was not hailed as the survey’s winner. It could be that the number of reported servers was not high enough to match AIX instances in the survey. But that’s conjecture on my part.
The reason why Linux as a class did not come in with less downtime than Unix is because the overall average downtime was made higher by reported Red Hat and Debian times: a (relatively) whopping 5.08 hours.
It wasn’t a total washout for Linux. As a category, Linux did have the most improved downtimes by far: Red Hat improved its downtimes from 2006 by 75 percent, SUSE improved by 73 percent, and even Debian made a 41 percent improvement.
Unsure About an Acronym or Term?
According to the survey author, analyst Laura DiDio, “In general, none of the major server operating systems — Linux, Macintosh, Windows or Unix — are today beset by the long list of bugs that plagued their predecessors back in the 1980s and 1990s.”
Well, that is certainly a valid statement. But not all of the server operating systems saw improvement. Windows operating systems actually had worse downtime stats in 2007 compared to 2006. Windows 2000 reported 9.86 hours (vs. 9.37 in 2006), and Windows Server 2003 went from 7.09 hours in 2006 to 8.90 hours of downtime in 2007. This was attributed to all of the critical updates to Windows that had to be installed in 2007. These numbers give the Windows category of servers only 99.9 percent reliability, a marked difference from AIX’s 99.999 percent.
The summary results of the survey are a fascinating read, as is the nifty [PDF] graph of the downtime results. Clearly, the Unix and Linux operating systems have made big strides in reliability improvements, no matter what your preferred flavor is.
If your cup of tea is Windows, however, its three-nines performance should make you stop and wonder about making a change.
Brian Proffitt is managing editor of JupiterWeb’s Linux/Open Source channel, which includes Linux Today, LinuxPlanet, and AllLinuxDevices.