Linux on desktop not cost-effective for most, says Gartner

Next week: analysts do it in harmony...

The second of this week's 'Windows cheaper than Linux' studies comes from Gartner, which tells us that "Linux on the desktop is not a cost-effective alternative for most enterprises looking for savings." That, however, is what the accompanying press release says - the Gartner research Macnugget the release is publicising doesn't put it quite like that, although it's certainly pointing in approximately that direction.

The authors argue that although "Linux has had significant success on the server", "the environment for Linux on the desktop is significantly different" (So far, so good, can't quarrel with that). But it goes on: "Knowledge workers use PCs to run diverse combinations of applications. For these users, migration costs will be very high, because all Windows applications must be replaced or rewritten.

"As a result, migrating desktops to Linux only makes sense in a very narrow, limited range of situations. The Linux migration should be considered only if there are relatively few applications, and these applications are fixed-function or low-function, such as data entry, call center or bank teller/platform automation. In these cases, the cost of migration may be low enough to justify the move to Linux."

Which, if you accept the basic premise, justifies the claim that Linux is not cost-effective for most businesses - but should you? The term "knowledge worker" has always seemed fairly dubious to The Register - it's a doubtful catch-all we strongly suspect Microsoft made up to help its marketing droids feel good about themselves, and we think maybe Gartner is overplaying the importance and numerical strength of the knowledge worker here.

Many people, in most companies, use the kind of narrow range of applications Gartner is suggesting as applicable to Linux migration here, and many use them in a "fixed-function or low-function" way. But this 'narrow range' includes Microsoft Office or similar.

At this juncture we invite you to run a free Register migration Total Cost of Ownership model we've just thought of. Consider whether what we're saying here is true for your company, whether what Gartner is saying is true, or if neither, where in between the two it lies. If a lot of people in your company run Office, but not in anything you'd describe as power user mode, then you'll tend to agree with us. If a lot of people run quite a number of high power or bespoke Windows apps, and/or have done a lot of tweaking and customisation, then you'll agree more with Gartner.

And actually both points of view are sort of right, and sort of wrong, because practically all companies will contain a mix of the two. So our beef with Gartner here is largely about where they decided to place the bar. No cheating when you run the Reg free TCO test, by the way - be cynical, and keep in mind that a lot of people who claim they're knowledge workers are only pushing Powerpoints back and forth as an alternative to having to think.

There are however a couple of obvious areas where this mightn't be the case. In the case of Office, real power users may have effectively locked themselves in, while companies with a fine and detailed macro collection that may be used on a widespread basis within the organisation might find they've locked the whole company in. Similarly, if they've done a lot of application development under Windows then they'll face costs in terms of tools, skills and development time to switch them over.

Gartner's claim that "all Windows applications must be replaced or rewritten" is however extreme, and in most cases wrong. When costing out a migration you'd first identify the areas where functionality can be duplicated fairly easily in Linux equivalents, then put a lot of thought into figuring out how you deal with the exceptions. You might do this by retaining some Microsoft machines, by using emulation to run whatever Windows apps, or by using a thin client approach. Or you might rewrite some apps, but surely not all of them.

And once you've worked out how many exceptions you've got, and how much it will cost to cater for them, you'll be in a better position to judge whether the move makes sense, and therefore where the bar lies in your specific case.

Gartner also has a TCO analysis, but you've got to pay for it so we're only able to share with you the information that Gartner reckons that a Win9x to Linux migration might make sense, and that this could save $80 a year in hardware acquisition costs (does this make Windows hardware?) plus $74 a year saving for StarOffice over Office. It also suggests that StarOffice on Windows could net you the same software savings, which is worth bearing in mind, but only works if you have Windows lock-in because of non-Office apps you're using. Its point that a "locked-down environment is key" to the achievement of savings under Linux is also kind of sensible, although might have been better put as it being easier to achieve savings through lock-down via Linux.

Really, the document is nearly sensible (not rocket science, sure, but what do you expect for free?). You can save money from migration, but you need to figure out whether you will or not, rather than just assuming. Gartner's problem, we think, is that it's following the official Microsoft line on Windows versus Linux TCO too closely, and that it probably needs to update itself on what the distributions are doing to tackle the classic gotchas (i.e. the exceptions as we define them above). ®

Related links:
Gartner report
Delay Linux apps, look at Windows, Unix, says Gartner
Linux in Munich - Gartner gets retaliation in prematurely?
MS' Linux obsession - time to call in the shrinks

Sponsored: 10 ways wire data helps conquer IT complexity