The Myths About Desktop Linux Persist, or, at least, that's the title of a Gartner Group "Managment Update" filed on April 13th of last year.
Here's the thumbnail:
The benefits of a Linux operating system (OS) in a desktop environment continue to be overhyped. Contributing to this misinformation are eight myths that Gartner identified in 2003, as well as two more recent examples that further exaggerate the advantages of deploying Linux on desktops.
The numbered statements in the list below are quoted from the Gartner work. The text below each is my attempt to summarize their argumen(s) for considering the numbered statement a myth and dismissing it as a reason to consider desktop Linux.
This is false because StarOffice and OpenOffice.org both run on Windows too, but aren't as complete as Microsoft Office. So there's no cost argument for Linux and anyway the tools are appropriate only for "users with a low reliance on Office".
No, you pay for the professional releases by paying for support.
There are - and they're more frequent than with Windows because Microsoft offers longer term support while the Linux community gets bored with old releases and moves on.
Gartner agrees that Linux will be more stable and less affected by PC style "security" issues, but argues that Windows offers better management tools. In summary, they say: "Gartner believes that significant reductions in staffing are not likely to be achieved simply by switching OSs without changing policies, lockdown or the degree of management tool implementation."
Gartner, however, thinks Linux more suitable than Windows for under-achieving sysadmins: "Although the Windows registry has helped give structure to Windows applications, it's hard to understand and repair. Since Linux is purely file-based, administrators may be able to troubleshoot application problems more easily."
Nope, Windows offers more and better lockdown and policy enforcement tools and most existing problems occur because companies fail to apply them.
Yes, but only for things that aren't important. For important stuff, while "StarOffice/OpenOffice.org is attempting to provide sufficient features and compatibility at a lower price point than the incumbent"..." Gartner has not seen this trend occur throughout the range of software that companies use."
While it's true that older PCs can run Linux with StarOffice, keeping them around introduces complexity into the lockdown environment, risks higher future Linux license fees, and forces the company to maintain PCs that could be junked and replaced for less than maintenance cost.
Yes, Unix skills from the data center transfer to Linux, but that's a negative because data center skills do not transfer to desktop support.
No, Microsoft EA licenses provide perpetual access to the latest technologies. Organizations which choose not to renew enterprise license agreements with Microsoft should "continue running the Microsoft products for as long as possible to maximize the value of their EA investments" -i.e. until sometime between 2009 and 2013 when Windows/XP and Office 2003 support was then scheduled to end.
Not true, "but there may be some users that can be moved. Look for large numbers of users running relatively few applications (browser-based, if possible) that do not heavily rely on Microsoft Office, no significant macros or Visual Basic for Applications (VBA), as potential targets for a partial Linux deployment."
Look past the author's thinly disguised contempt for Linux as a kind of second rate Windows 9X substitute, and the message seems to be clear: companies considering desktop Linux should consider all desktop software; be realistic about costs; "understand that locking down and managing the desktop will result in bigger savings than simply switching operating systems"; and, "consider technical, political and cultural issues to assess how desktop management practices may succeed under Linux, if they previously failed under Windows."
On the surface that might strike you and me as fair, but remember that Gartner's market is the technically illiterate senior manager - and that person is going to get the clear message that companies suffering Wintel upgrade and security traumas have incompetent IT staff, with anyone suggesting Linux as a way out obviously among the worst of the worst.
To put it nicely this isn't the most honest bit of analysis I've read today, but it's hard to disagree with the real bottom line here: that client-server is client-server, and neither its costs nor its limitations are much affected by a move from Windows to Linux on the clients.
Since, in my opinion at least, the problem with Windows client-server is client-server, not Windows I thought it might be fun to weigh Gartner's claimed Linux myths against a smart display architecture implemented using Solaris on SPARC with Sun Rays.
Remember, in a smart display architecture there is no processing of any kind on the client - it's not a thin client architecture: it's a DoIP (display over IP) architecture featuring big, fast, and utterly reliable desktops, centralized processing, and decentralized management.
And, incidently, if you think the smart display architecture is the mainframe come again, please read the Gartner bit above about managing PCs via lockdown and policy enforcement again -that's the mainframe come again. Take an an honest look at the smart display alternative and you'll find it's the opposite: easily the smartest dumb idea around.
Please note, too, that I've had to adjust Gartner's wording a bit for the text below to make sense - with my changes inside square brackets. So:
This is true, but somewhat irrelevant because the Sun Ray can provide Microsoft Windows and Office access concurrently with Unix access - so you can have Word on XP and Swriter on Solaris open at the same time and on the same screen if you want to pay for the MS licenses needed to do it.
True - 10 year old software (and ten year old smart displays) runs unchanged under Solaris 10 on the UltraSPARC T1 or IV+ today.
True for many applications. Some per user licenses are significantly more for Solaris, but they can usually be shared among multiple part time users to produce a lower cost per user.
True. My orginal dual 167Mhz Ultra2, now upgraded to a magnificent 2 x 296Mhz, is currently running a production application under Solaris 10.
True (but not actually relevant to anything)
Bottom line? Gartner puts up 10 straw men and shoots them down as arguments for using desktop Linux. The whole thing is largely specious with some serious wordsmithing and elision deployed to make sure the ten myths can be debunked as stated, but there is a kernel of truth to their underlying argument: that switching desktop operating systems without switching how those systems are used is something of an excerise in futility.
Switch desktop architecture, hwoever, and all ten "myths" turn out to be perfectly reasonable arguments for abandoning, not Microsoft Windows, but Microsoft client-server.