Reliability: Unix and Linux beat Windows (heise online) [LWN.net]

Reliability: Unix and Linux beat Windows (heise online)

Posted Apr 18, 2008 6:07 UTC (Fri) by kripkenstein (subscriber, #43281)
Parent article: Reliability: Unix and Linux beat Windows (heise online)

I wasn't surprised to see Linux outperforming Windows, but what *did* surprise me was Ubuntu
besting Debian.

Since Ubuntu is based on Debian unstable, and much more fast-moving than Debian in general, I
always assumed Ubuntu would be less reliable. But in this survey it had ~1 hour downtime vs.
~5 for Debian.

Can anyone explain this? Random thoughts of mine include that Ubuntu has commercial support
(so people end up using it more correctly perhaps), or maybe that Ubuntu issues patches after
more testing. But I have no idea, this is surprising to me.

(Log in to post comments)

Reliability: Unix and Linux beat Windows (heise online)

Posted Apr 18, 2008 6:34 UTC (Fri) by jordip (subscriber, #47356) [Link]

The answer should be the people using those systems. 
For what I read around Internet and my own friends, Debian administrators tend to be more
hackish than Ubuntu's or other distro. 
So they tend to have more resources solving problems but on the other hand they may prefer
their own methods over what the distro have ( or what the distro doesn't have and they
implement).
As everybody is wrong sometimes, they will make the system fail more often that people that
just use what the distro provides that at least has been tested by someone else than yourself.

Don't get me wrong, even if those systems have a couple more bugs they have to address, some
of them are pure art in what they are achieving.

Debian reliability on the report

Posted Apr 18, 2008 22:50 UTC (Fri) by man_ls (subscriber, #15091) [Link]

The answer lies more in the atrocious reporting methodology. Centering just on one sentence, the relevant slide (a ridiculous pdf with 2 pages) says literally "OpenSource Linux (e.g. Debian)". Where:

"OpenSource" does not exist: according to its proponents it is "open source".
Most Linii (or Linuxes, if you prefer) on the page are "open source": Red Hat, SUSE (not "SuSe" as it appears, which is not even "SuSE" as in the original German), Mandriva, Ubuntu, etc. No reason to single out Debian.
"Open source Linux", even if referring to "Community distro", should include everything from Fedora to Debian to Gentoo to home-grown distros.
Even if limited to distros maintained by the community, there are a lot of them including Debian, Gentoo, Puppy Linux...
Even if limited to Debian, the report makes clear that "24% of the respondents reporting they had at least one Debian server in their network", which does not look like actual production servers. Rather it might be experimental or unmaintained machines.
Debian administrators are not "hackish". Depending on how you define downtime, something which is of course absent from the report or the slides, Debian servers do not suffer downtime just because of bugs; once they are in production they stay there for years. What is true is that Debian administrators may work in more precarious conditions than others: I used to run several experimental servers and my only downtimes were related to grid failures, something which a proper setup would have avoided.
And as pointed out above, Laura DiDio is not likely to elicit honest responses from a variety of Debian administrators.

There are many other problems with the report. Just see how "Other Linux" varies wildly from one year to the next, or how last year's "Unix (AIX, Solaris, HPUX)" with 6,54 hours of downtime have magically transmogrified into three categories with less than 2 hours each. I seriously doubt proprietary Unix has turned 4 times as reliable in this time. And "Customizations" have mixed effect on reliability, even worse when combined with last year's data.

It is therefore hard to extract meaningful conclusions from the figures reported. If I had to, I'd say that people are still reluctant to run mission-critical services on Debian, and that is why Debian servers are more likely to suffer downtime than the Red Hat counterparts. In this case it is no reflection on the quality of the OS, but on how much the machines are cared for. And that is maybe why proprietary Unices have grown in reliability: only legacy services are still running on these servers, and they are the most reliable since they do little work and are never modified. More demanding applications have migrated to Linux in many cases.

Reliability: Unix and Linux beat Windows (heise online)

Posted May 3, 2008 18:24 UTC (Sat) by anton (guest, #25547) [Link]

what *did* surprise me was Ubuntu besting Debian.

My explanation is: A company that pays for round-the-clock staff usually will want a commercial enterprise distro installed, not a community project like Debian. So the larger downtimes of Debian systems compared to others are not due to a higher failure rate, but due to a larger average downtime when a failure occurs; e.g., if the system crashes on Friday night, it may have between several hours and 2.5 days of downtime ahead. Yes, a hardware watchdog would help availability, but for some servers availability is not everything.

Since Ubuntu is based on Debian unstable, and much more fast-moving than Debian in general, I always assumed Ubuntu would be less reliable.

Ubuntu does its own testing, so the difference may not be so big.