| About InfoWorld : Advertise : Subscribe : Contact Us : Awards : Events : Store |
 |
||||
|
||||
 | ||||
 |
|
|||
 |
||||
|
|
|
||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||
 TEST CENTER The future of Windows By Tom Yager , InfoWorld Test Center March 14, 2001 Will Microsoft conquer the corporate world with its new recipe for an enterprise application architecture? The answer depends on whether or not .NET is capable of competing with Java
Judging from IT buzz on the subject, business doesn't find Microsoft's latest server OS all that compelling either. There is, as yet, no great corporate stampede to upgrade Windows NT 4.0 servers to Windows 2000. Some businesses are waiting for .NET to take shape. Others are postponing big IT purchases until an economic upturn, lower interest rates, lower memory prices, 64-bit Intel CPUs, or whatever else the tea leaves tell them is a sign to move ahead. Plenty of prophets are weighing in on the lukewarm response to enterprise Windows. The most outspoken of the Linux camp have touted the new 2.4 Kernel as the enterprise Windows killer. Sun and Oracle have foretold the fall of Windows and the failure of the PC. The U.S. Department of Justice has tried to weaken the giant by cleaving it in two. In the end, none of these factors will have a significant impact on Windows' large-scale penetration, either in the near term or the long term. Windows' prospects for enterprise success rest solely on Microsoft's ability to convince the corporate market that .NET, Microsoft's application and integration strategy, somehow meets enterprise needs better than Sun's Java. It's the architecture, stupid During Microsoft's humble period in the wake of the Justice Department's antitrust suit, a company executive was widely quoted as admitting that Windows NT 4.0 had serious reliability problems. But these days, Microsoft's marketing campaigns brashly tout Windows 2000 as being "five nines" (99.999 percent) reliable. Microsoft is not the first software company to underplay the shortcomings of its products. Everybody but Sun seemed to know that Solaris 2.4 was the buggiest Unix ever made, and wasn't the 2.2 Linux Kernel supposed to give the enterprise everything it needed? Even so, selling Windows 2000 on its superior reliability doesn't wash. There are too many rock-solid Windows NT server installations, and no Unix adherents are going to switch to Windows 2000 because they are told it's more stable. Microsoft should apply the same reliability standard to its message as it does to its server OS. By lifting the veil on .NET before either the technology or the message had congealed, Microsoft scared the hell out of some of its staunchest supporters. If my company is now heavily invested in Windows Distributed interNet Architecture (DNA), Microsoft's current EAI (enterprise application integration) architecture, will that investment be lost when .NET appears? Will .NET get my disparate enterprise applications talking to one another, a job that Windows DNA never mastered? Finally, am I making a big mistake by committing my scarce dollars and resources to new projects based on Windows DNA? The .NET message has wandered significantly, but one theme has been constant: .NET is a new architecture. It is the progeny of Windows DNA, but you can't take full advantage of .NET without rewriting your applications. Period. Microsoft is trying to stoke its sales by stamping the .NET moniker on every server product it rolls out. Servers running .NET applications will need SQL Server, BizTalk Server, some selection of other server facilities, and a server edition of Windows. Although these pieces are available now, the essential .NET underpinnings -- the object model, the run-time engine, the APIs -- are a long way off. No combination of currently available products delivers on the promise of .NET. The .NET architecture does not replace the COM (Component Object Model) architecture, the meat and potatoes of Windows DNA. The two models will run side by side for many, many years. But the commercial delivery of the .NET architecture, coupled with the key enterprise applications that use it, will turn COM into a legacy technology. Windows DNA applications work too hard to exchange information and functionality with one another. The .NET model is infinitely cleaner and simpler. Windows DNA is a patchwork of technologies that were woven together while Microsoft was still figuring out what an enterprise application is. Sun had the luxury of defining Java 2 Enterprise Edition (J2EE) after all that work had been done. In developing .NET, Microsoft was able to create a recipe for an enterprise application architecture from scratch. That is .NET's strength and the key to its ability to compete with J2EE in the enterprise. The Windows XP experience There isn't much news to report on Windows' server side since Windows 2000 rolled out last February. This February, Jim Allchin, vice president of Microsoft's platforms group, and Bill Gates, Microsoft's chairman and chief software architect, ended months of speculation by officially yanking the veil off Windows XP and recasting it as a consumer product. If that's what it is, this is the kindest turn Microsoft has done consumers since the introduction of Windows 95. We understand Microsoft's desire to reserve the Windows 2000 brand for its corporate product line, but we think Windows XP has several significant advantages over Windows 2000 Professional and Windows Me for corporate desktops, notebooks, and remote clients. First, Windows XP finally unloads the baggage of DOS. Windows Me keeps DOS under wraps, but Windows XP is Microsoft's first consumer OS to dump this legacy millstone entirely. DOS has hung around because common maintenance activities such as BIOS upgrades and file system repairs required it, but hardware and software vendors have finally heeded warnings about DOS's demise and built 32-bit alternatives to DOS utilities. For those legacy applications that simply won't run under 32-bit Windows, a new application compatibility feature in Windows XP spins a cocoon that tricks the application into thinking it's running under an older OS. Windows XP improves the user experience in important ways. Even systems not joined to a Windows domain support multiple users, and application state is maintained when a user logs out. When users log back in, their open applications are restored. The Windows XP GUI takes advantage of ubiquitous display acceleration hardware. It uses wide color palettes, animation, and transparency to make Windows interfaces visually appealing, without wasting pixels. Two features make Windows XP a hands-down winner in the corporate space. The OS ships with an authenticated remote control feature, allowing a user in trouble to send an e-mail plea for assistance to the help desk or another employee who can provide assistance. After the message goes out, the user's system listens for a remote control attempt from the invited Samaritan. The feature that may attract the most attention is the mandatory, "phone home" Internet registration. That's right: Microsoft's honor-based licensing system is history. Following installation, each protected product must be connected to the Internet long enough to exchange registration data with a Microsoft server. This will drive some people nuts (see our Review on Microsoft Office XP), but it's a good thing for business. Microsoft currently places responsibility for license management in customers' hands, but if every licensable product must be activated via the Internet, responsibility shifts to Microsoft. If Microsoft wants support for its new licensing model, it must give up its right to examine paper licenses on demand. The fact that an electronically activated software product is installed must constitute proof that the product is properly licensed. The future of Windows in corporate server rooms, on desktops, and in attaché cases hinges on issues that have seemingly little to do with operating systems. A server OS is judged mostly for its facilitation of a workable enterprise application architecture. Windows does that today to the extent that the market requires. For cross-platform applications, Windows shops can run J2EE on Windows servers. For Windows enterprise applications, they can boost performance by using Microsoft Windows DNA. Looking ahead, businesses should focus their enterprise Windows efforts on applications and infrastructure components that fully embrace .NET. Related articles Installers ease app distribution pains Office spruced with surprising subtlety Visual Basic.NET makes demands on businesses Microsoft's Allchin extols five benefits of .NET Users size up Web services Developers must focus on linking Web services to boost Net collaboration Web services Tom Yager is the East Coast technical director of the InfoWorld Test Center. His views on enterprise software come from almost 20 years in corporate IT. Send your views via e-mail to tom_yager@infoworld.com.
 RELATED SUBJECTS  Operating SystemsSPONSORED WHITE PAPERS 
SPONSORED LINKS
|
|||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||
|
||||||||||
