Competition is always healthy

As the Microsoft antitrust case nears its conclusion and the dust from Comdex settles, you might understandably think, after glancing at the trade weeklies or listening to a sound byte from a presidential candidate, that Windows is about to lose substantial market share to Linux. Don't believe everything you hear.

Yes, Linux will gain momentum as a server OS, particularly with ISPs. And Corel's recent introduction of Linux for the desktop and planned release this year of six office productivity applications will increase Linux's desktop presence in academia and staunchly anti-Microsoft shops. As the Macintosh has, Linux will garner a small but loyal following, but the OS still lacks the broad third-party hardware and software support necessary for mainstream acceptance.

Linux is also attempting to make inroads into the thin-client market. In the Windows 2000 Magazine Lab, we're testing a large group of thin clients for an upcoming comparative review. As we selected products for this story, we found three that use embedded versions of Linux and one that runs an embedded version of Windows NT 4.0. The Linux-based units were middle-of-the-road performers in our tests, and they were more difficult to set up than the NT and Windows CE units were.

Without realizing it, Corel and other vendors that have worked to make Linux more competitive might have spurred Microsoft to make Windows 2000 (Win2K) a better product. A little competition never hurts—look at how AMD lit a fire under Intel. We've heard about several UNIX shops that are considering migrating to Win2K, and other companies might follow this lead. Certainly, most current NT shops don't seem anxious to defect. In a study that Windows NT Magazine and World Research conducted in May 1999, 92.5 percent of the study's 1439 participants said they plan to eventually deploy Windows 2000 Professional (Win2K Pro) and Windows 2000 Server (Win2K Server). More than 80 percent of the study's participants said they will deploy Win2K Pro and Win2K Server by the end of 2000.

At press time, Win2K's release to manufacturing (RTM) is still a few weeks away, but our experience with the most recent release candidate (RC) indicates that Win2K will address many of NT 4.0's shortcomings. Even so, some companies will wait until the OS has been available for a while before they evaluate its performance and reliability to see whether these factors offset Win2K's increased hardware requirements and deployment costs.

The Lab's policy is to not publish performance test scores on prerelease versions of a new OS, so we've been waiting for the gold code's release before we run extensive benchmark tests on Win2K. However, Microsoft's Win2K product management team offered to meet with Lab staff a few weeks ago to discuss their performance tests, and we accepted the offer. The Microsoft team based the data they presented to us on testing they conducted with a variety of benchmarks on RC2 and RC3. The Microsoft team claims that these results should be virtually identical to results we might obtain with Win2K's release versions. The Microsoft results showed substantial performance improvements over NT 4.0 in file, print, Web, and application server tests. The workstation tests showed that Win2K Pro is slightly faster than NT Workstation for running high-end applications on powerful workstations, and roughly equal to NT Workstation for running mainstream business applications, even on a 133MHz Pentium processor with 64MB of RAM.

If Win2K embodies these performance claims and the OS's reliability improves, then IT managers will be amply rewarded for their Win2K deployment efforts. However, we in the Lab will reserve judgment until we can validate Microsoft's results with our own tests on Win2K release versions. We don't always believe what we hear.