Windows Bumps Unix as Top Server OS

Ivan writes " Windows narrowly bumped Unix in 2005 to claim the top spot in server sales for the first time, according to a new report from IDC. Computer makers sold $17.7 billion worth of Windows servers worldwide in 2005 compared with $17.5 billion in Unix servers, IDC analyst Matthew Eastwood said of the firm's latest Server Tracker market share report. "It's the first time Unix was not top overall since before the Tracker started in 1996.""

Linux?

[Rob T Firefly]

From TFA:

And in another first, fast-growing Linux took third place, bumping machines with IBM's mainframe operating system, z/OS. Linux server sales grew from $4.3 billion in 2004 to $5.3 billion in 2005, while mainframes dropped from $5.7 billion to $4.8 billion over the same period, Eastwood said.

"Sales" being the operative word. How would one fit the free Linux options into this equation, I wonder?

I have some numbers...

[spottedkangaroo]

We purchased five brand new Dell rackmountable servers last month. When we got them, we burned in some linux and threw the windows disks in the trash...

Comment removed

[account_deleted]

Comment removed based on user account deletion

Re:GNU

[WhiteWolf666]

Slight marketshare loss for Unix, large marketshare growth for Linux, with Windows edging out Unix minus Linux.

Another thing that is sometimes forgotten

[Anonymous Coward]

Linux servers often outperform Windows servers so you can have only one or two linux servers where you would need 5 windows servers. I know it's anecdotal, but I think others can back me up one this. Linux can get more done with more stability. However, Windows server 2003 certainly has been a huge improvement over previous Microsoft server OSs. If I was in the market I would definitely not choose what OS to use simply because it's Microsoft or simply because it's open source. I would choose the OS that suited my needs and the abilities of my employed system administrators.

It's nice to have two products which are pretty much equivalent in performance. I think this is the type of situation which is good for competition, and good for everyone involved. Even if Microsoft does take the lead in server marketshare, I think that's just naturally reflecting the skillset of the workforce. Many people who are becoming system administrators now only know how to use windows, and probably haven't even used dos.

Re:Article seems misleading

[gameguy1957]

I agree.

We have to run a seperate server for each app that requires Windows as the server. So, instead of one server to run the apps we have four.

Our Linux servers do multiple duties. Same goes for the Novell servers we use. They all perform multiple tasks on top of the standard filesharing and print q's.

We had to pay for each copy of Windows server, Linux was free to install anywhere and the Novell is a site license per student so we can install it as many times as we want without additional fees.

What I'd like to see is the number of new installs in the past year. Win vs. Unix vs. Linux vs. others.

-JM

Check my math...

[Anonymous Coward]

Well if what I am reading is correct Unix is $17.5 Billion + Linux $5.3 Billion = $22.8 Billion. Microsoft is $17.7 Billion. Last I checked the *nix camp is still $5 billion ahead of M$. Not to mention that there was no number on Mac OSX in there. Just another feel good bullshit article trying to pump up M$.

Re:Servers

[Syberghost]

What about server hardware sold without an operating system?

It's counted. They're using a combination of methods, remember, and that includes asking those surveyed "how many servers did you buy without an operating system, and what operating system did you put on them?"

Absolutely true, apples to oranges...

[tizzyD]

It is incredibly difficult to produce a "market" leader measure without some consideration to the way that the market is measured. Fundamentally, that method determines the leader. Consider the obvious:

• If you measure by units sold in a particular month, you would miss a sizable chunk of the market. How? Well, for nearly every 4 years, every system I installed at SMBs (small to mid-size businesses) was either Debian or Gentoo. As such, there was no direct cost associated with those units, but they were the foundations of many applications, file and print services, email services, directory services, databases, etc. More and more, as much of the functionality of a file and print server is commoditized, it can be handled more cost effectively by Linux, Samba, and other OS apps. Therefore, since these items incur no revenue in the market to a company, they would not be counted. Thus the distortion.
• If you consider units deployed, you have a difficult data mining challenge. How do you collect the vast amounts of data? As a researcher at times, you'd have to subscribe a number of organizations--we're talking hundreds--and then over the span of years, see what their deployment considerations are. From that measure, you can more accurately determine the statistically valid (within 5% perhaps) measure of deployed systems, more accurately demonstrating a market. It's a market, but in a different way, that is, for ancilliary products and services, upgrades, etc.
• If you measure a market by sales, you distort the market by not considering all forms of distributed products. When I install a MS system, there typically is required a number of ancilliary products that must be installed, including things like SQL Server (to hold the LDAP store). Are these sales counted as part of the market? Without Active Directory, you almost can't do anything else--SharePoint, Exchange, etc. Therefore, it is almost a component of the OS. On a comparable *nix, you would simply use a compliant LDAP system, but then, you would not consider it part of the OS. Considering the LDAP may be from another company, it further distorts the true market.

The market measure should be considered a dubious statistic, much like a political one. Raising the overall spending on education means nothing. Raising the overall spending per student, that means something. If you raise overall spending per student in constant dollars (inflation adjusted dollars), now you are really producing an accurate measure. The fact that most people can't understand basic comparisons--read the book Innumeracy [amazon.com] by John Allen Paulos--leads to this fallacy of a measurement.

Re:How long

[zootm]

How about the 12 servers we bought from dell without an OS that had linux installed on them? I know those were not counted. Or the 100+ servers acting as mpeg2 video routers in the headends that were also bought without an OS that has linux installed as well.

Linux was listed in the ranking seperately (it came third, according to the article). Linux is not UNIX, so even if you bought it with Linux installed, it wouldn't have changed the relative positions of the two operating systems.

I agree that Linux would have come higher if the eventual OS installs of servers with no OS installs had been recorded, though.

Wrong, sorry...

[killmenow]

It would be pretty stupid to lump UNIX and Linux sales together, given that Linux is not UNIX. As far as I can tell, not a single Linux distribution is certified against the Single UNIX Specification [unix.org], which any Operating System must be in order to be UNIX.

They share similarities to be sure, but they are not the same and should not be lumped together any more than Windows and Linux should be lumped together.

Re:How long

[notaprguy]

Unix servers on average cost significantly more than Windows servers. Even educated people seem to have trouble getting it through their heads that the cost of the cost of the software is not the most significant cost in setting up a server. Also, I think you're right that you'll see more Windows servers but it's probably not because of more teaching facilities focusing on ASP.NET and the like. It's because IT decision makers (the people with the $$) are going with Windows for a whole variety of reasons (works for them, good TCO, good integration with the rest of their stack etc.)

Re:Free servers

[HaydnH]

Actually they aren't giving them away, they're letting you trial them for 60 days. Some customers who create a comparative report (to another os or chipset server) might get to keep their server - most will have to buy or return it, from sun.com:

"Thank you for your interest in our Try and Buy Offer. For a limited time, Sun is offering qualified customers a free 60-day trial of the world's first eco-responsible server, the Sun Fire T2000 server. And, it's easy...Your complete responses allow us to instantly determine your qualification and get your trial system to you within a couple of weeks! It will only take about 3-4 minutes to determine your qualification. After the evaluation period, you may choose to purchase the server or return it to us."

Although they do look like nice servers - I wouldn't want to reutrn it ;P

Re:How long

[zootm]

Linux is not Unix, it's "Unix-like". I'm not particularly anal about that distinction, but there is one, and the distinction was made in this test. Solaris is "real Unix" (as I believe AIX is, I'm not sure about the BSDs). There's a specification for "what Unix is", so as nit-picky as it sounds, there's a technical reason that they don't count. There's a bit more explanation on this Wikipedia page [wikipedia.org].

I was just trying to explain why Linux wasn't counted in the Unix ranking. On the other hand (as another reply to my post has pointed out) Linux, rather than Windows, is likely to be the reason that "real Unix" is losing market share.

Re:Sheer number of small servers

[Anonymous Coward]

Depending on what agreements your company has with Microsoft that may not be the case.

I don't know about your particular situation, but my experience is that most corporate agreements with Microsoft apply only to the desktop licensing rather than server licensing (in most cases, desktops outnumber servers by a good amount). Licensing server operating systems per vm instance is a real cost to most companies.

Jim

PS: I work at a company where we have over 100,000 users (presumably a similar number of desktop systems with a large number of servers) and our agreement only covers the desktop licenses for the site. Any servers purchased must be purchased with the OS license rather than a corporate agreement.

Re:How long

[Mr Z]

Well, yes and no... Various Linux distros have angled for POSIX certification and certification against the Single UNIX Specification, [wikipedia.org] and so on, and I believe some have actually made it, though I don't see anything over at the Open Group's site.

Barring a huge merge between Linux and a "real UNIX," Linux will never be a "real UNIX" based on source code descendency. But, Linux may be considered a "real UNIX" at points due to SUS certification of particular distro releases that care enough to bother.

And, as far as BSD is concerned, while it may have removed certain proprietary AT&T code as part of the lawsuit, enough code and structure crossed both ways between BSD and AT&T that it'd be silly to argue that BSD is not "real UNIX." That said, it appears none of the BSDs have registered for UNIX certification. [opengroup.org]

--Joe

Re:How long

[smoker2]

But basically, *nix servers outsold windows servers by $5.1 billion (that's UNIX and LINUX combined).

So I guess there is a little breathing space yet ;-)

Interestingly, if you RTFA and scroll down to the other links, you'll see "windows leads server OS pack for first time" last november ! [com.com]

dupe or astroturf - you decide ...

Re:How long

[joelleo]

Its a matter of ease of deployment and use. Most likely, this little law office isn't going to have an IT guy and most of the day-to-day will fall on the office manager, which most likely doesn't know SaMBa from doing the Rumba. In that situation, the cheapy solution makes absolutely no sense whatsoever.

With the addition of a single server running Small Business edition, includes a bunch of useful stuff - Exchange, ISA, SQL etc, they can have a _supported_ solution covering their current needs and many/most of their near-future needs. Plus, with the addition of AD, reduce the amount of effort that goes in to managing the day-to-day stuff. Instead of managing 11 machines' passwords, she manages the domain. This includes homogenizing the configuration of all the workstation and all the other benefits of having centralized control of configuration, authorization and authentication. Workgroup isn't a "solution" in any sense :)

Some will argue that SBS is harder to manage for the general user wearing the admin cape - true, but how much easier is it than trying to manage a linux server for them?

Re:How long

[Zeinfeld]

"A law office with ten employees using Windows XP is going to buy a Windows server, end of story." Why would they do that? This is actually an ideal place to implement a Samba solution. At 12 systems they aren't even looking at a domain, they are looking for a workgroup solution. There are advantages in cost, reliably, and all aspects of performance.

Because there is a lot more to Windows server than NFS support. There is no open source system that provides a turnkey replacement for active directory.

The cost of Windows server is irrelevant compared to the cost of having someone set up any enterprise class server. IT consultants charge upwards of $2000 a day.