I'm posting a comment that came into my personal blog. Jon makes a great point about what the CIO knows vs. what his/her employees are doing with Linux:
"
The other question I have, which will probably just be embarrassing to the CIOs interviewed would be, how many of those shops already use Linux and are planning on deploying more without the CIO even knowing it is happening. It’s hard to imagine a modern enterprise shop that isn’t already using Linux and already planning on deploying more appliances, servers, and virtual machines with Linux underneath.

In my experience CIOs don’t really know what is driving their applications. They operate with budgets and projects. Totally results driven, due to who they report to, and how they justify budgets.

Linux on commodity Intel hardware is a very capable replacement in the server room for proprietary UNIX and Windows platforms, and data centers which embrace the solution only seem to flourish.

The only time I see an enterprise shop make a move off of Linux in favor of something more expensive, is when a vendor sweetens the pot. This happens more than you would think, but it creates a love-hate relationship with said vendor.

Cheers,
Jon
"

I agree completely. Many CIOs only make decisions that set directions or standards not actual purchases or specific implementations. These same CIOs will be getting their data for recommendations from their staff. Many of those directions are influenced by perks given by vendors (free golf, dinners, etc...). I would be willing to bet money on the facts that more than 50% of CIOs have no idea what platforms they are running the software on.

An interesting survey question for the mid-level people would be "Are you prevented from using Linux due to company decisions or policies"

These 'analysts' drive me nuts. Not so long ago, someone from TheStreet said that "Microsoft was cleaning with Windows", but they look at revenue, not units (Linux is set up for free and it's cheaper... up over 30% in the last quarter... 34% IIRC). So never mind if companies like Google build a 1-million unit global datacenter with Linux and never pay $3000 for some Microsoft servers. Linux is getting extremely eay to set up and there's no licensing issue... no paper trail.

Never mind, the analysts just look at money, money, money... and spin figures their own way, sometimes to please someone that pays them.

By the way, Barron's got caught shilling for Microsoft several times recently. I reported this as well.

Roy Schestowitz wrote: "Never mind, the analysts just look at money, money, money... and spin figures their own way, sometimes to please someone that pays them."

In defense of analysts everywhere, Heather is a financial analyst and unless you are Swiss, a druglord, another very rich person, another large bank or a pension fund you are unlikely to be dealing with her firm, UBS. It is unlikely that a gnome got up yesterday morning and told her to put out those statistics to move a little Microsoft stock out of UBS' portfolio or to lower the price of Red Hat further.

As a market researcher, the real problem with UBS' methodology, if I read the Infoweek article correctly, is that the firm only interviewed 60 firms. So all those 60%'s and 90%'s represent just 30-50 guys.

Instead, see my analysis of the most recent related IDC numbers at "http://seekingalpha.com/article/46357-microsoft-nearing-complete-dominance-of-the-server-market." But don't look at them if you're a Linux bigot. Unfortunately for you, IDC's statistically significant server tracking research, run quarterly, says the same thing as UBS' unscientific research.

It is fairly well established (I'll provide references if you require them) that IDC and InformationWeek are very close to Microsoft. Their figures, 'studies', and deliberate omissions were identified before.

Please do not resort to using stereotypes like "Linux bigot". I'm a scientist. I am just sick and tired of seeing statistics being bent to support financial interests and align with some hypothesis that suits those who buy 'studies'.

There is already enough corruption in the world (see OOXML vote buying and deceitful press releases). We don't need more of it.

Roy Schestowitz wrote "It is fairly well established (I'll provide references if you require them) that IDC and InformationWeek are very close to Microsoft. Their figures, 'studies', and deliberate omissions were identified before."

Roy, sure send in your references or post them and I'll take a look. But I dispute that it's "fairly well established" that all the disparate sources you attack are in the bag for Microsoft: IDG's IDC, UBS' financial analysts such as Heather Bellini (original research but methodologically flawed as noted in my post above), thestreet.com, Dow Jones' (soon Newscorp's) Barrons, CMP's Information Week (probably not original research since the latter three are news organizations), and so forth.

I am not sure what kind of scientist you are but I hope you are one that looks at the numbers dispassionately. Of course, studies are almost always skewed one way or the other. And you can argue (as I get paid to do) about methodologies and taxonomies. But that is not the same as the statistics themselves: the numbers are the numbers. Umpires just call 'em as they see 'em.

As an aside, I meant the you after "Linux bigot" in the plural and was not referring to you in particular. Sorry if it offended you. But I would say you are a Linux bigot. That's nothing to be so senstitive about. For example, I'm a Coca-Cola bigot; can't stand Pepsi even though Coke and Pepsi are probably 99.999% the same.