Will Longhorn be worth the pain?

Microsoft has removed some of the functionality from the next version of Windows to bring forward the release date, which could make the decision to upgrade even more difficult for some companies

After re-jigging the 'three pillars' of functionality that underpin the next version of Windows, codenamed Longhorn, Microsoft will need to work hard in the months ahead if it is to provide customers with compelling reasons to migrate.

The three pillars of the software giant's next generation Windows OS, due to ship in late 2006, originally comprised Avalon, a new graphics and presentation subsystem, Indigo, a Web services and communications architecture, and WinFS, a new file system.

But Microsoft has now decided to backport Avalon and Indigo to Windows XP and Windows Server 2003 to encourage developers to start building applications, which means the features are no longer unique to Longhorn. It has also stripped out WinFS completely and will not now make the offering available in the initial client release of Longhorn at all, let alone the server version. Full details of the three pillars of Longhorn, and details on what has happenned to them since they were first announced, can be be found on the last page of this article.

Neil Macehiter, a partner at MWD Advisors, believes that the move reflects the importance that Microsoft attaches to companies signing up to its Software Assurance programme — a three-year software maintenance scheme by which mid-sized and large enterprise customers agree to pay a set annual fee to receive upgrades without having to pay extra. The obvious appeal of the scheme to Microsoft is that it receives a predictable revenue stream.

But customers, especially in Europe, have been voicing dissatisfaction because many feel they have not received value for money over recent years from the SA scheme. This, according to Annette Jump, a principal analyst at Gartner, has led the software vendor to try and "add value through initiatives such as training and support to persuade companies to sign up again and in some cases, it's even reducing prices".

But Macehiter believes that Microsoft's back-porting strategy, while useful to encourage crucial developer buy-in, is also a "double-edged sword".

"Microsoft must provide developers with access to the technology — hence the availability of Avalon, Indigo and WinFS on Windows XP and Server 2003 — but that means the applications will also be available on pre-Longhorn releases of the OS, potentially reducing the incentive to upgrade," he says.

As a result, the decision to "carefully extract" the three, much-hyped pillars "leaves Longhorn on shaky foundations and Microsoft will have to muster all of its marketing smarts to highlight the benefits of the under-hyped operating system fundamentals such as security and management capabilities".

Jump likewise feels that the removal of WinFS in particular means that the client version of Longhorn at least is unlikely to be as "revolutionary" a release as first expected.

"WinFS was very compelling, but Longhorn is still changing every six months so it's a bit early to say whether it will be very different to XP. We'll just have to wait and see what's included, but it appears to be just another Windows OS that's more of an evolution than was originally planned," she says.

Mark Quirk, head of technology at Microsoft's developer and platform group, acknowledges that making Avalon and Indigo available for Windows XP SP2 and Windows Server 2003 SP1 means that "the core of Longhorn is an evolution of where we are today".

"We do see the usability and productivity characteristics of Longhorn as more revolutionary, however. We know computers are still hard to use and manage and we expect Longhorn to deliver major improvements here," he adds.

While Quirk refuses to be drawn on specific functionality, saying that Microsoft doesn't "want to give the game away too early", he does hint that more will be revealed when the vendor ships the first Longhorn beta, which is expected some time this summer.

He also indicates that the supplier's key marketing messages now relate to increased reliability, security, and usability, which, he claims, will result in higher levels of end user productivity and "empowerment".

Another key catchphrase, "ease of deployment", will also be highlighted. "We're doing a lot of work on deployment both for new customers and upgrades. We think it's one of the most fundamental areas to address and we'll have new facilities in Longhorn to make it easier to upgrade," Quirk says.

And one compelling reason to upgrade, he believes, is Longhorn's new Active Protection feature, which recognises the presence of viruses or worms by their behavioural characteristics rather than their signatures and is also aware of whether a given machine is patched or not.

But Olivier Nguyen van Tan, chief operation officer of Pierre Audoin Consultants, believes that Microsoft will also push a fifth key tag line in the form of 'integration innovation'.

"When Microsoft started its platform strategy about three years ago, it was the first time that it had pushed the idea of integration innovation, meaning that all of its products are linked together and work together efficiently. Integration is a big pitch for Microsoft because the appeal to the market is that it means more productivity and less cost," he says.

But the big question, of course, is whether the messages around such apparently mundane, albeit useful, functionality will be enough to tempt the average customer to switch, particularly given the time and resources that such upgrades demand.

Jump warns that even a minor OS upgrade generally takes enterprises between nine and 15 months to complete, including testing and roll out, while major releases can take anything from between 12 to 18 months.

Over the next two years, organisations should try "as much as possible" to move their PCs to the latest version of Windows, which in the case of the client version means XP, because it will make migration to Longhorn much easier, explains the Gartner analyst.

Although it is currently unclear how much of a leap from XP to Longhorn it will be, Jump does expect the Avalon user interface element to make the move as tricky as it was from Windows 3.1 to 95, which involved the rewriting of many older applications.

"If you skip versions and aren't on XP, you'll have to do a big bang migration and you need a very big budget for that. Also it costs more in the long-term for a big migration," she warns.

The issue is, however, that only 32.5 percent of the total desktop installed base is currently using XP, while another 38 percent is on Windows 2000. A further 2 percent employ MacOS, 1.5 percent have Linux clients and the remaining 26 percent still use unsupported versions of Windows, including 95, NT 4 and Millennium Edition.

In short, if nearly two thirds of Microsoft's customers need to move to XP before they even consider migrating to Longhorn, it appears that the supplier has quite a major challenge ahead.

On the upside though, Quirk indicates that the OS will run on the same hardware as XP works on now, which will at least save money here, although he "won't promise that it will run on a low spec machine".

But the downside for Microsoft is that none of the analysts are advising organisations to rush into anything, recommending rather that they take a pragmatic approach and adopt the OS only where they see real business value.

"The key is to focus on the value of the upgrade. Applications that provide compelling features may make it worth moving, but you also have to think about how much it costs to do mass deployments," Macehiter concludes. "There are no hard and fast rules, so I think it will be a trickle migration and consumers will probably take it up faster than companies because they'll buy new PCs with Longhorn loaded."

On the final page, you can find an explanation of the the original 'three pillars of Longhorn', and what happenned to them in between their announcement and the current time.

The Three Pillars of Longhorn

Avalon
Avalon is the code name for a variety of presentation class libraries that will be deployed as a common front end for all newly-developed Windows client applications, documents and graphics, which can be represented in either two or three dimensions.

It supports the XAML, which is similar in nature to XML and enables developers to compile code on the fly. It also means that user interfaces can be written separately from the underlying business logic.

Avalon will appear for use with Longhorn at the same time as the client version of the operating system in the second half of 2006 or slightly before, but will also now be made available as a download from Microsoft's Web site for Windows XP and Windows Server 2003. The first Community Technology Preview was released last November and an update is due to appear this month.

Indigo
Indigo is the code name for a set of class libraries that provide the communications infrastructure to allow different .Net components to talk to each other. It combines support for Web services development; Microsoft's COM+ object model; MSMQ messaging; peer-to-peer communications, and .Net remoting (enabling client applications to use objects found in other processes) in one environment.

Indigo will appear for use with Longhorn at the same time as the client version of the operating system or slightly before, but will also now be made available as a download from Microsoft's Web site for Windows XP and Windows Server 2003. The first Community Technology Preview was released earlier this month.


Avalon + Indigo = WinFX
Avalon and Indigo taken together comprise WinFX, Microsoft's new object-oriented application programming interface, which the vendor is positioning as a superset of the .Net framework.


WinFS
WinFS is the code name for a new unified Windows storage subsystem, which is intended to enable users to undertake searches based on the metadata of any stored item, no matter what type of file format it has or which type of application created it.

The technology was originally supposed to appear in the client version of Longhorn, but will now only be available in beta release in this timescale. It will also ship for Windows XP, but not for Windows Server 2003. This is consistent with former Microsoft announcements that there would delays to the server version of WinFS.

But as to what the technology will eventually look like in either client or server form is currently unclear as the vendor refuses to commit to firm details.

WinFS has been compared to the much-derided Cairo version of Windows, which was announced in 1992 and supposed to include a new user interface and Object File System data store to store files and documents in any format, but never made the light of day. The technology is expected to emerge in some form, however, although it currently appears to be somewhat of a moveable feast.