The long road to Longhorn

Microsoft Corp. is developing versions of its Windows operating system with only a subset of the Windows code base, designed for specific server tasks, in a move that could reduce maintenance costs for customers and create products that are less vulnerable to attack.

The new “role-based” products may appear in 2007, when the server version of Longhorn is scheduled for release. Offering a smaller code base would mark a significant technical shift for Microsoft and could help it to better address the competitive threat posed by Linux. But it also presents significant engineering challenges for the company, industry analysts said.

Microsoft already sells two role-based versions of Windows – one for storage and one for Web serving. They essentially conceal from users the parts of Windows that are not needed for the task at hand, making them easier to install and use. But those products still are based on the entire Windows code base.

With its Longhorn release, Microsoft hopes to offer role-based versions of Windows for tasks such as storage, file, or print serving that include only a part of the overall Windows code, said Martin Taylor, Microsoft general manager for platform strategy. “Today, it’s still the entire code base. There’s no reduction in the bits you get; things are just roped off,” Taylor said. “We want to get to a model of role-based deployment where you might just have the bits you need for that function…. It’s one of our design goals for Longhorn.”

Many enterprises separate tasks such as e-mail and Web hosting, assigning them to individual servers or groups of servers. Microsoft would continue to sell the full version of Windows for use as a general purpose operating system, but could offer the role-based products for servers assigned to particular tasks, he said.

Such a move could benefit customers in two primary ways: better security and lower maintenance costs, said Michael Cherry, lead analyst at Directions on Microsoft Inc., a research company in Kirkland, Wash.

Reducing the amount of code on a server would reduce the “attack surface area,” Cherry said, meaning hackers would have less code to aim at with their viruses. A virus last month that targeted a component for viewing JPEG image files affected Windows Server 2003, he noted, even though customers managing their servers remotely don’t need that component on their servers.

Having less code should also mean lower maintenance costs, in part because customers will not have to apply patches to the parts of Windows that do not exist on their servers. “If you have a server whose role is definable and distinct, and you only load enough (code) to carry out that role, then it’s got to bring the total cost of ownership down,” Cherry said.

A Microsoft spokeswoman confirmed that the goals of providing a smaller Windows “footprint” are to cut maintenance costs and provide a “reduced surface attack area.”

Taylor didn’t say whether the role-based server editions would be lower in price than the full versions of Windows. The role-based products it sells today – Windows Storage Server 2003 and Windows Server 2003 Web Edition – are cheaper than the full versions of Windows Server 2003, although usage restrictions apply.

Removing parts of Windows presents engineering challenges. For starters, Microsoft must ensure that ISVs (independent software vendors) can continue to write applications for its operating system without worrying about which components are available to them, Taylor said.

“The nice thing about Windows today is that Windows is Windows,” he said.

Microsoft is also keen not to limit the software’s management capabilities. For example, the current Web Edition of Windows cannot serve as a domain controller, which means it can’t be used to manage group policies, Internet authentication services and other tasks.

“You have trade-offs in cost and trade-offs in manageability, so it’s a fine balance,” Taylor said.

He did not provide an exact timetable for the products, but said development work is “pretty far down the path.”

Microsoft can predict, broadly speaking, what roles its customers want their server software to fill, but it can’t predict every combination, Taylor said. For example, a customer might want to install a print server and a Web server on one system, and a storage server on another. He suggested that Microsoft may engineer the products in a way that allows customers to decide at deployment time which parts of the operating system they want to use.

“We can predict the roles, but some customers might want Web and print on the same server,” Taylor said. “Instead of delivering these preconfigured, maybe there’s a way to do it with a bit more customer input.”

Some industry insiders have speculated that Microsoft would reduce the code base for certain server versions of Windows. When it developed Windows XP Embedded several years ago, part of the task was to figure out which parts of the operating system could be removed without disturbing the software’s internal dependencies, Cherry noted.

“That knowledge of how the components work together could allow them to build versions of Windows with only specific functionality…. It could be used to develop several other products,” Cherry said. The proposed changes to Windows come as Microsoft faces perhaps its toughest challenger yet in the form of Linux. The open source operating system has been gaining traction among governments and businesses, in part because some view it as a cheaper and more secure alternative.

One advantage of the Linux source code being publicly available is that customers – or more commonly systems integrators or resellers – can take the kernel apart and use only the components they want, to build a DNS (Domain Name System) server, for example, said Neil Macehiter, a research director at U.K. analyst company Ovum.

“I think they need to move to this model of having role- or function-based servers to compete better with Linux,” he said.

SIDEBAR: Users react to Longhorn moves

By Carol Sliwa

Computerworld (U.S.)

The next major Windows release, code-named Longhorn, is due to arrive so far into the future that many corporate users aren’t particularly bothered by Microsoft Corp.’s plan to scale back the product.

Prior to its late-August announcement, the company had pledged that Longhorn would feature three innovative new elements: the Avalon graphics subsystem, boasting special 3-D effects; the Indigo communications subsystem for building advanced Web services; and the WinFS storage subsystem, which aims to provide advanced data organization and management capabilities and make it easier to search for information.

But in order to hit its targets for delivering the Longhorn client in 2006 and the server operating system in 2007, Microsoft said it won’t include WinFS. The storage subsystem will be in beta testing when the Longhorn client is released, the company noted.

“It isn’t that big of a deal,” said Roger Wilding, a Portland, Ore.-based senior technical engineer at shipping and supply chain services company CNF Inc. “It would be nice to have (WinFS) in the next release of Windows for the desktop, but I would prefer a clean, stable product.”

Victor Stuart, an Indianapolis-based business solutions analyst at Sallie Mae Inc., said the change in plans for Longhorn won’t affect his company at all. “We don’t plan on using things that either don’t exist or are merely part of a planned future release,” Stuart said.

Rob Rhodes, a technical consultant at Louisville, Ky.-based Kindred Healthcare Inc., said the decision to remove WinFS from Longhorn sounded reasonable, since he thought Microsoft was “biting off quite a bit.” Rhodes said he even likes the idea of WinFS potentially being an add-on to the operating system, since that would give his company the option of not using the technology if it adversely affects hardware requirements or performance.

Greg Sullivan, lead product manager on Microsoft’s Windows client team, said WinFS is being developed independently of the client and server operating systems and will be delivered separately. He said that when finished, WinFS could be applied and installed on one of the existing operating systems.

“WinFS is their ticket for next-generation searching, and it also contains new APIs that could potentially compete against the Java platform. So there’s more riding on this component than there is on Avalon and Indigo,” said a software engineer at a major pharmaceutical company who asked not to be named. “If something’s going to drop off the first release on Longhorn to make sure they get it right, it should be this.”

But some users would prefer WinFS over other new features. Jason Glazier, senior vice-president and chief technology and e-commerce officer at Lincoln Financial Group in Philadelphia, said WinFS was “one of the more significant advantages” of Longhorn. He said its removal “definitely detracts from the value of the Longhorn release and makes it less likely we will move when it is released.”

Walt Smith, chief architect at a U.S.-based financial institution, said that he would like a more robust server environment as soon as possible and that 2007 “seems a bit far off for serious planning purposes.”

He added that he would gladly forego Avalon’s 3-D graphics in favor of WinFS. “It’s unfortunate that Microsoft has chosen to delay the delivery of potentially important improvements in the way that information is managed in the enterprise in favour of delivering more glitz and glamour in the front-end graphics subsystem of Longhorn,” Smith said.

Sullivan said there will be more new features in Longhorn beyond those that Microsoft laid out at last October’s Professional Developers Conference. He said Microsoft has yet to detail new functionality that will appeal to business users and consumers.

But Thomas Groves, a systems engineer at Oregon State University in Corvallis, said WinFS is the feature that interested him most, with its promise of greater file system speed, less-frequent file system degradation, improved data retrieval and seamless integration between network and local data.

“WinFS has the greatest potential to revolutionize Windows,” Groves said.

Related Download
3 reasons why Hyperconverged is the cost-efficient, simplified infrastructure for the modern data center Sponsor: Lenovo
3 reasons why Hyperconverged is the cost-efficient, simplified infrastructure for the modern data center
Find out how Hyperconverged systems can help you meet the challenges of the modern IT department. Click here to find out more.
Register Now