Unix & Linux Stack Exchange is a question and answer site for users of Linux, FreeBSD and other Un*x-like operating systems. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Please check figure 2 in this page: Linux versus Windows: OS impact on uptime and speed

enter image description here

Why is NetBSD the most stable OS here?

share|improve this question

closed as primarily opinion-based by jasonwryan, slm, Anthon, msw, goldilocks Aug 24 '13 at 5:11

Many good questions generate some degree of opinion based on expert experience, but answers to this question will tend to be almost entirely based on opinions, rather than facts, references, or specific expertise.If this question can be reworded to fit the rules in the help center, please edit the question.

This must be a fake. Websites run by Minix systems. I wonder why they haven't included CP/M. – ott-- Aug 23 '13 at 18:36
Perhaps NetBSD is being used not as a general-purpose server OS, but as a specialized embedded OS in a special-purpose appliance, such as a load balancer. – 200_success Aug 23 '13 at 18:48
@ott : Nice catch. It's that or a total botch up, I think, as, the link claims 6.4% of the 12,000+ sites surveyed were using minix o_O – goldilocks Aug 23 '13 at 19:48
The link is from 2007 btw. – On a eu. Aug 23 '13 at 20:07
@goldilocks Or in other terms, it found Minix to be about 15% more common than Linux in serving web sites. Tanenbaum called; he wants his monolithic kernel back. – Michael Kjörling Aug 23 '13 at 22:01

I would call in to question the experimenters methods here.

Next we analyzed the uptime and response speed per each Operating System. We are monitoring every 30min simultaneously from 3 locations in the USA, Germany, and Austria, so we are performing 1,008 checks per week per each website. That means for example that we made for detected Linux servers 7,295*1,008 = 7,353,360 checks in week 16. Similarly we did 2052*1008= 2,068,416 checks for Windows servers in the same week. That gives us a solid base for OS to OS comparison, assuming that we may neglect other differences (e.g. in each group there will be similar per cent of good ISPs, good application designs, etc.). When we calculate averages and sort data by uptime we got the following pictures (per week):

It seems like there could be drastic problems here. Perhaps a server rebooted in the 30 minute window. Perhaps another server was seen as being down once even though it took 59 minutes of downtime (not 30 or 60). They did have a fairly high sample size though. In reality, most production websites are up considerably more than 99% of the time. There may have been just a couple "bad apples" in each bunch of operating system.

In short, I consider those results to be approximately accurate, but unreliable.

Edit: uptime does not necessarily equate to reliability either. It could be that the admin is also the dev and does not have a dedicated dev machine. As mentioned in the blogpost, the ISP could be flaky. Really, this test was not conducted scientifically at all. These results are novel at best.

share|improve this answer

Even if they were otherwise methodologically perfect — which they are not — you cannot compare sample sizes of 7295 of 238 and derive anything meaningful from their relative means.

That's not how statistics works.

share|improve this answer

You would have to take a much closer look at the servers being monitored to make an intelligent judgement, the numbers are very close.

You can get five nines from any one of the above mentioned OS. The question really is how hard are you going have to work to maintain five nines and what risks will you take to do it.

Windows typically has allot of security/other bugs and an SA would be considered negligent not to install fixes that often require reboot.

Netbsd has a very low number security/other bugs thus can go for longer between reboots before SA bumps up against due diligence.

As for response time, NetBSD systems are often optimized systems that have one function such as DNS, other network functions, or even firmware etc. So hard to tell what is going on just by mentioning the OS.

share|improve this answer

I would just like to add the fact that Linux offers a much wider and better support on the driver-side, NetBSD, as all the major BSD OS, is a great OS, but most of the time I still use it as a "toy" because I just don't get the necessary drivers with it, and you can feel this gap across the board, from the VGA to the ACPI functionality, there are a lot of peripherals that simply work better, and are more stable, under Linux.

There are also huge differences in the release cycle, weeks for Linux, months/years for BSD-like OSs, whether you consider this a good thing or a bad thing it's likely to influence your final choice.

share|improve this answer
How does this answer the question? How is it even related to the question? (Except for kernel updates, most updates on Linux can be installed with little more than a service restart which might take a few seconds, and they aren't measuring uptime anyway.) – Michael Kjörling Aug 23 '13 at 21:46
@MichaelKjörling the question is about stability and I'm taking this on the driver and kernel side. – user1154 Aug 23 '13 at 21:53
The fact that you don't get the hardware support you need in any particular OS doesn't mean it is in any way less stable on some other hardware. This answer seems to be more an anecdote about your experience with running BSDs, than an answer to the question of why NetBSD would be found the most stable OS by the particular study cited, which didn't even measure uptime but rather some sort of "up-ness" -- and found Minix to be about 15% more common than Linux on the Internet (not likely). – Michael Kjörling Aug 23 '13 at 21:58

Not the answer you're looking for? Browse other questions tagged or ask your own question.