ONLamp.com    
 Published on ONLamp.com (http://www.onlamp.com/)
 See this if you're having trouble printing code examples


ONLamp 2005 Survey Results

by chromatic
12/02/2005

We recently ran our annual ONLamp.com survey to find out more about you, our readers. As usual, we learned many interesting things, not just about you but also about our questions and our survey itself. 587 people completed the survey, which is enough to get some good statistical analysis. Here are some of the gems.

Demographics

Seventeen percent of our readers visit the site daily, with 49% visiting at least once a week. Great! Of the 3% who said "other," most free-form comments were essentially "when the RSS feed shows me something interesting." There's probably some overlap here. We don't have a lot of good statistics about traffic that comes through syndication and feeds, but there's plenty of it, as later questions reveal.

Ninety-seven percent of readers read the articles and 44% read weblogs. We publish a lot of weblogs -- what can we do to make them more interesting? Which topic areas would you like us to cover in more detail there? Did you know that each author and topic has a dedicated feed?

Eighty-one percent of readers prefer programming tutorials. 69% enjoy power user tips and hacks. 60% like system adminsitration information. The second seems surprising, but we can improve our power user coverage (especially for the F/OSS desktops). What do you consider a power user in this context?

Related Reading

Linux Server Hacks, Volume Two
Tips & Tools for Connecting, Monitoring, and Troubleshooting
By William von Hagen, Brian K. Jones

Fifty-seven percent of readers navigate our site by browsing the front page, 15% use an outside search engine, 38% use a feed reader, and 24% find stories through the Linux newsletter. Perhaps there's a selection bias here, in that the people who read the Linux newsletter were more likely to hear about and to take the survey. Several of the "other" responses included links from other sites. We like getting referrals.

ONLamp.com was the first place 5% of readers had heard of O'Reilly Media, while 11% weren't sure. Those results seem pretty sound to me. We don't do a lot of publishing that appeals to people who aren't already in technical circles. We could do more -- but it seems more likely that you'll pass along links to interested-but-not-early-adopter technical friends and family, especially for our "What Is" articles.

ONLamp.com ranks favorably among our readers' favorite sites, being the favorite for 34% of respondents and among the favorites for 52%. That's wonderful! Among the other favorite sites, Slashdot and Newsforge top the list with 82% and 37% of visitors. LWN has 16% of visitors, newcomer LXer has 7%, and the comments section included too many other sites to mention more than a few: Perl Monks, other O'Reilly Network sites, Linux Today, the Linux Journal, Artima, Undeadly.org, and IBM Developer Works. Obviously people understood this as "favorite site for the topics covered."

Almost half (49%) of our readers work for small businesses of 1 to 50 people. It's important to keep small business in mind, especially in a world that seems to goggle slack-jawed in awe at anyone who drops the word "enterprise" into polite conversation. (You should perhaps instead view such people with pity, shock, and mild distaste.) Fifteen percent work in large companies of over 2,500 people, with 13% in businesses of 101 to 500 people, and 9% each in 501 to 2,500 and 51 to 100. Five percent don't know, which actually makes sense when you get above a couple of dozen but remain below a thousand.

Twenty percent of readers work in the computer software industry, with 12% in internet commerce, 11% in education, 7% in networking and telecommunications, and 5% in government of any size. "Other" answers included science, biotechnology, research, students, and self-employed consultants.

Thirty-seven percent of readers have 3 to 5 years of practical "Linux" (there's that word again) experience. 20% have 6 to 8 years and 18% have over 8 years. Wow. Thirty-four percent of readers have 9 to 19 years of professional experience related to their current positions, 25% have 6 to 8 years, 18% have 3 to 5 years, and 12% have over 20 years. You're an experienced crowd.

Seventeen percent of readers consider themselves primarily software developers and programmers (with 8% calling themselves software engineers, a sign we ought to merge the two next time). Thirteen percent are application developers, 12% system administrators, and 7% webmasters. There are plenty of students and researchers in the 12% who answered "other."

The bulk of our readers (46%) are between 25 and 34 years old; 27% are between 35 and 44, with 12% being 18 to 24, 11% being 45 to 54, and 3% being older.

Nearly half (49%) of respondents live in the United States, 32% in Europe (a big "country"; we should break this down further), and 6% in Canada. Nat Torkington just increased the number of readers from New Zealand by 10%.

Ninety-five percent of our readers are male.

Technical Questions

Several of our questions used the word "Linux" where it's more accurate to say "Linux, a BSD, or another open source operating system." We'll try to correct that for next year. The phrasing this year skews the results somewhat, with some respondents commenting "I don't use Linux" or "I use BSD but counted it as Linux." It's difficult to get good demographic data sometimes when you walk the line between being too specific for one group (the people analyzing the data) and not specific enough for another (the people answering the questions). Even if we did change the wording for next year, the lines are still fuzzy. Does Mac OS X count as an open source operating system? Darwin does, but Aqua and Cocoa don't -- but what if you run primarily open source applications on it? What if you primarily run Cygwin and remote X11 applications on a Windows machine? It's still a fuzzy question.

With that in mind, 12% of readers use Linux at home only and 13% use it at work only. That seems a bit backwards, if it's even statistically significant, but the fuzziness of the possible answers doesn't help. Sixty-two percent use Linux both at home and at work and 12% don't use Linux.

Our readers use plenty of technologies, including multiple operating systems: 83% use Linux, 62% Windows, 55% Unix, a respectable 37% BSD, and 33% Mac OS X. There are plenty of databases too: 71% use MySQL, 34% PostgreSQL, 21% Oracle, and 59% SQL in general.

Few readers are monoglots. Sixty-one percent use PHP, 55% use Perl (hey, we have a website about that too!), 47% use C or C++ (we ought to separate those for the next survey), 37% use Python, 21% use Ruby (moving up in the world), 16% use .NET, and 8% use Mono. Other technologies include Apache (presumably Apache httpd) at 78%.

For development systems, 72% of readers use Linux, 55% Windows, 25% BSD, 22% Mac OS X, and 13% Solaris. It's likely that some people use Unix boxes for development, though they tend to use Windows machines for desktops. (My most recent development job was much more pleasant after running a remote X session from a Linux machine on Cygwin's X server and ignoring as much of Windows as possible.) In the comments section, a few people reported AIX, a couple VMS, a few IOS and VxWorks, and one DG/UX. Wow.

Looking at the systems our readers deploy to may help to confirm that: 81% deploy to Linux, 51% to Windows, 31% BSD, 18% Solaris, and 18% Mac OS X. Correlating these answers to the types of software developed might be very interesting. There are also a few deployments to AIX and HP-UX.

In terms of tools used, 59% of readers use gcc, 50% CVS, and 40% Subversion. It'll be nice when those last two numbers reverse -- perhaps next year. Other standout tools were Eclipse (29% of readers) and gdb (28%). In the comments, vim came up often, as did emacs, Ruby on Rails, Komodo, and Visual Studio. One of those things is not like the other (hint: Ruby on Rails).

Twenty-one percent of readers consider Debian GNU/Linux their primary distribution, 17% Fedora Core, 14% Red Hat, 12% SuSE, 10% Ubuntu, and 8% Gentoo. Mandrake had a small showing, but JDS, Conectiva, and TurboLinux did not. Many of the comments mentioned FreeBSD and a few Slackware. (One of the admins here swears by Slackware.)

Our next question was quite bad, asking how many CPUs readers have in their largest Linux system. Obviously (now), saying "one to five" is fairly ambiguous, if we want to know if you're doing SMP. We'll fix or flush this question for next time.

Eighty-four percent of readers run primarily open source software on servers, with 58% running it on developer systems, 46% on administrator systems, 31% on desktops, 13% on application terminals, and 8% on thin clients. 7% don't use it anywhere. 76% of readers plan to migrate to an open platform for at least one category there, and of those who do, 61% plan to migrate servers, 41% developer systems, 38% administrator systems, 36% desktop systems, 17% application terminals, and 15% thin clients. Thin clients are on the rise, but servers, developer systems, and administrator systems remain the domain of open source software.

As alluded to before, our readers mostly use MySQL, with 57% of those answering the question mentioning it. Eighteen percent use PostgreSQL, 13% Oracle, 2% SQLite, and 1% Sybase. In the comments, plenty also use SQL Server, a few Firebird, and the others you might expect. Users who plan to switch databases prefer PostgreSQL (49%), then MySQL (26%), Oracle (11%), with 3% SQLite and 1 Sybase. Not one percent, one reader. In the comments, Firebird came up most often.

Plans and Predictions

Around one-third of respondents are taking concrete steps to move users from Windows or Macintosh systems to Linux. The comments indicated that we worded this question poorly -- some saw no reason to switch, others moved to Windows and Mac OS X from Linux, and some migrated to BSD. Several respondents are ISPs or web companies where it makes little sense to talk about migrating customer desktops.

Of web developers, 49% are planning to deploy web services -- defined as a SOAP or REST interface -- on their LAMP systems. My favorite comment was "CGI isn't an interface anymore?"

Predictions are always fun . . . of the topics our readers expected to make it big in the next few months, there was plenty of Ruby, lots of Ruby on Rails, quite a few Ruby on Rails competitors and workalikes, and more than a few responses of "Perl 6." Keep the faith!

chromatic manages Onyx Neon Press, an independent publisher.


Return to ONLamp.com.

Copyright © 2009 O'Reilly Media, Inc.