Commentary: State of the server OS wars

The week of June 20 saw interesting tidings in the x86-based operating system front--good news for Windows enthusiasts, and bad news for the Linux contingent, says Enterprise columnist Alan Zeichick.

The week of June 20 saw interesting tidings in the x86-based operating system front--good news for Windows enthusiasts, and bad news for the Linux contingent.

First, the Linux news: LynuxWorks, a Linux vendor that specializes in developing embedded operating systems (systems that go inside Internet appliances, hand-held computers, and industrial equipment) decided to postpone its initial public stock offering. Remember how the spectacular IPOs of Red Hat and VA Linux propelled Linux to the front pages of daily newspapers?

How times have changed. This is the fourth Linux IPO to be abandoned over the past few months, following LinuxCare, TurboLinux, and Lineo. One would be right to question the viability of the for-profit Linux business model.

On the Windows front, Windows 2000 Server got a boost at Microsoft's Tech-Ed conference in Atlanta, where Bill Gates unveiled a new name for the next-generation "Whistler" server operating system: Windows .Net Server.

Microsoft is stuffing a lot of new functionality into its server software, including an improved middleware layer that supports distributed messaging, the .NET framework for Web services, a revamped infrastructure for serving dynamic Web applications, and a common language runtime environment. Microsoft hints that Windows .Net Server might be generally available in the first half of 2002.

Score one for Microsoft: Linux has no native middleware equivalent. The only way to add those features to Linux is to buy an application server and messaging queue software based on Sun's Java 2 Enterprise Edition (J2EE). Unless you go with an open-source implementation of an application server or a message queue, a Linux plus J2EE per-server solution will cost far more than Windows: You're looking at tens of thousands of dollars per processor.

Ouch! And even there, Java's Web services features won't be standardized until J2EE 1.4, expected in fall 2002 or in early 2003. Any Web services functionality would be based on proprietary implementations from BEA Systems, HP Bluestone, IBM, IONA, iPlanet, Oracle, or another J2EE vendor.

Does this mean that it's game over for Linux? No way, José.

First of all, few enterprises are rushing to implement Web services. The whole concept of Web services is immature. Even though some of the underlying technologies are established standards, such as XML, SOAP (Simple Object Access Protocol) and WSDL (Web Service Description Language), many developers are worried about security, reliability, security, performance, security, scalability, security, and pricing.

With few exceptions (such as Microsoft and a handful of partners), nobody's deploying Web services. There's plenty of time for the Linux/Java community to catch up.

Second, even if Microsoft has a more coherent and compelling distributed-computing vision today, it doesn't have to stay that way. Sun's Java specifications include a top-notch message-queuing specification, Java Message Service (JMS), which is every bit as good as Microsoft's COM+. The problem is, JMS is currently available only in the expensive J2EE specification.

If Sun were to move JMS from J2EE to the freely available Java 2 Standard Edition, Java could give Windows .Net a run for its money not only on Linux, but also on Unix and other platforms. There's been no such announcement to date, but privately some members of Sun's Java Community Process are strongly urging such a move.

Newsletters

You have been successfully signed up. To sign up for more newsletters or to manage your account, visit the Newsletter Subscription Center.
See All
See All