Microsoft, Sun Do Well in Open Source Census

Astute observers of U.S. presidential elections know that at 12:01 am on the first Tuesday of November in every leap year, a small group gathers at the Balsams Resort Hotel in Dixville Notch, NH, a few miles south of the Canadian border, for their quadrennial 15 minutes of fame. They vote and go back to bed, certain that all the initial election-day news stories about “early returns” are really talking about this totally unrepresentative group of grizzled “Yankees*”.

What’s this have to do with information technology investment research? No, this is not another rant about the crazy Diebold (DBD) voting machine conspirators on the open source software fringe.

This is about the OpenLogic Open Source Census, which includes good news for both Microsoft (MSFT) and Sun (JAVA). I have been a big fan of OpenLogic’s idea since meeting with Kim Weins, Marketing VP, and Steven L. Grandchamp, CEO, in October 2007. Kim gave us a preview of what they were doing a few months later in a podcast. (I would even like to see the idea go further and become a true census of all types of software, which would end the useless “us vs. them, Open source vs. Microsoft” thing altogether.)

But the OpenLogic Open Source Census is in its Dixville Notch stage. As of September 30, slightly more than 2000 “machines” have been scanned, since the census began in April 2008, discovering over 700 “brands” of open source software deployed in 300,000 discrete open source software instances.

For Microsoft (a census sponsor), the good news is that much of this open source software is running on Windows. As of an earlier snapshot in August 2008, more than 50% of the open source software was Windows based but that metric was not released this time (see earlier parenthetical phrase). I suspect that the number has dipped below 50% but the prevalence of Samba as the 19th most frequently found package tells me the Windows number is still high. To me that means that open source is driving business for Microsoft even if its shareholders believe that Unix-Linux migrations should have been Unix-Windows migrations.

For Sun, the good news is the popularity of OpenOffice; it’s on 45% of the machines scanned. Remember, those most likely include Windows machines. But that’s good news of course only if Sun can turn OpenOffice users into StarOffice subscription users or otherwise monetize the popularity of the Office competitor that Sun made open source back in 2001. MySQL is also found on 28% of the machines scanned.

Ruby, a growing software package not associated with any major public company, also scores well as we come out of the hills of New Hampshire.

Surprising against conventional wisdom is the low prevalence of open source software middleware such as Mule, Red Hat (RHT) JBoss and the IBM (IBM)-led Geronimo project. That’s another indication that we are in the Dixville Notch, non-representative stage of the census. I believe the initial census results reflect development vs. deployment machines in a manner unrepresentative of the real world.

Going forward, one of the problems with the census is its opt-in nature. That provides all the big software suppliers a way to cook the books but at least it’s an equal-opportunity temptation and they all know how to play the game. Over time, if the mix of machines represented in the OpenLogic survey approximates the aggregate IDC quarterly data on PC and server shipments, it should start to truly represent reality. I would guess that by this time in 2009, all the votes will be counted.

(*For non-U.S. and even non-New-England and even non-Vermont readers who want to know what a “Yankee” is, see this link, particularly E.B. White’s definition within the Wikipedia entry.)

Comments

[Josh Thompson:]

How is it determined if the software was installed but not used. That is, I have software on my PC that I installed in order to try out the software and decided it was not for me. I didn't bother deleting it. This would be more common if the network admin installed the software and I couldn't delete the software even if I wanted to delete it.

2008 Oct 02 10:34 AM

[The Grinner:]

Gee, Microsoft funded a survey that didn't include Mac OS X...what a crock. But then, Microsoft funded surveys have always been a joke, and never really mirror real life.

2008 Oct 02 12:22 PM

[Dennis Byron:]

The above commenters may not completely understand how the census works. The group conducting the census welcomes information from Mac, BSD, Solaris, and other platforms.

As for Java, I believe it is a prerequisite of even beginning the census so by definition it is on every "machine" scannned.

More information see https://osscensus.org/discover... .

Or just rant away!!

-- Dennis

2008 Oct 02 04:04 PM

[KimWeins:]

A couple of clarificaitions:

Microsoft didn't found the Open Source Census. We started it at OpenLogic and now have 13 additional co-sponsors. The co-sponsors include open source organizations as well as companies. Jim Jagielski of the Apache Foundation is an advisor. Microsoft joined as a sponsor later and wasn't involved in discussions or decisions on how to conduct the Census.

We take data for the Census from anyone that wants to participate -- including Mac users. If you check out the website at osscensus.org, you will see data from Mac users. We didn't quote it in the press releases because we don't have enough data from that group yet. We encourage more Mac users to participate!

As far as finding Java, we would need to add a "fingerprint" for Java. There is an open source community for the OSS Discovery tool that is used by the Open Source Census, and we take contributions for fingerprints from anyone that wants to participate. However, there is also good data available on Java adoption from other sources.

2008 Oct 03 10:31 AM

[Yo Ma Ma:]

Until they start using the GPL, there *is* no Microsoft open source.

2008 Oct 20 10:59 AM