I'm tired of reading on an almost monthly basis articles asserting that Linux isn't yet ready for the desktop. Nonsense! Linux is about as ready for the desktop as Windows is. It's simply a matter of corporate and user inertia that's keeping Linux marginalized.
There are really two markets for desktop Linux. One is individual users. I'm ignoring them here to address IT departments. If you're a Linux user outside of an organization, you probably face many of the same issues as corporate users, but you can (and must) deal with them yourself, and therefore decide for yourself whether Linux or Windows or MacOS suits your needs best.
The other market for desktop Linux is business organizations. Their OS needs are well-known: reliability, security, compatibility with existing data and devices, and the availability of applications they need to accomplish their business objectives.
To compete on the desktop, Linux must score at least as well on most of those criteria as the other choices, and it does. It's at least as reliable and secure as Windows XP. Microsoft might have the edge in compatibility with existing data and peripherals, but Linux has the edge in compatibility with existing clients -- you may have to buy a whole new PC to upgrade from Windows 98 or 2000 Professional to XP.
Application availability can be a killer. While there's now a greater range of open source applications for users, the number still doesn't stack up to what's available for Windows. But more important than what's available in shrinkwrap is what's been written for special purposes. While simple Visual Basic for Applications macros may run in OpenOffice, the highly customized ones that many businesses rely on probably won't. The same is true of applications programmed in high-level languages. Virtually any such program will require more than a simple recompile of the source code to work under a new operating system.
There's one other key component that I thought was hindering general deployment of Linux in large organizations -- lack of a good desktop management client for server-based network management frameworks. Applications such as LANDesk Software's LANDesk Management Suite, Veritas Desktop Management Suite, Novell ZENworks for Desktops, Computer Associates Unicenter, Tivoli, and Novadigm Radia Management Suite are vital to keeping costs under control when managing large numbers of clients. These programs must be able to perform hardware and software inventory, software distribution, remote control, and desktop policy management.
As soon as I did a little research, however, I learned I was wrong. According to a recent buyer's guide in Network World, leading products like Unicenter, LANDesk, and Radia already support Linux clients. It seems reasonable to assume others will follow suit.
Yes, there are other issues for organizations that switch operating systems, including training for users and system administrators. You certainly have to figure those costs into the decision whether to change desktop operating systems. But that doesn't negate the assertion that Linux is ready for desktop deployment now.
Of course, just because Linux is ready for the desktop doesn't mean users are ready for Linux. Users get very comfortable with their computers, and quite naturally fear change. But new systems come into production all the time as the needs of a business change, and when new client operating system better fit business needs, employees have to change with the times. The best way to assuage users' fears is with a phased-in approach, migrating one department at a time (starting with the most technically savvy) and learning from each move. Also, don't take away every Windows PC at once; leave one or two available in each department, if no longer officially supported, for those that might have a legitimate need for them.
Any network manager facing a desktop upgrade decision ought to factor Linux into his or her calculations. I haven't seen any hard figures (can anyone point me to any?) but it wouldn't surprise me if it took just as much effort to upgrade Windows 98 clients to Windows XP as it would to migrate them to Linux. The initial expense is less, and the total cost of ownership on an ongoing basis is lower. What's not to like?