Plan for a slow roll-out of Windows Vista

Delays to the Vista operating system should not prove a big problem for users as its immediate benefits may not make a compelling case for a speedy upgrade.


Microsoft's decision to delay the full release of Windows Vista should emphasise the importance of a cautious approach to the new operating system. Many of its features will be improvements on its predecessor, but users may not find them compelling enough to justify the cost of upgrading.

As with Windows XP and Windows 2000, Windows Vista will offer incremental, evolutionary improvements. Most of those  that directly benefit users are security-related. However, most of this functionality is already available via third-party products.

Windows 2000 users should plan to begin migrating in early 2008. Most Windows XP users should pursue a strategy of managed diversity, bringing in Windows Vista on new machines beginning in 2008.

With the release of Vista expected in less than a year, Microsoft's marketing machine is in motion. While many of Vista's features will be improvements on XP, XP is a good operating system that will be supported until at least October 2013. Microsoft's biggest challenge is to provide enough benefits in Vista to make users want the upgrade.

As PC hard drives have ballooned and e-mail traffic has increased, it has become increasingly difficult for people to find things. Search is slow in Windows XP, and files, e-mail and calendar objects cannot be found with a single search.

Most people still organise documents much as they did under Dos 20 years ago - by folder, based on the application they were created in. Because it owns the user interface, Microsoft has the potential to do a more seamless job of integrating search into the user navigation experience with Windows Vista.

But there are other ways to improve searching. Competent third-party desktop search tools are already available for Outlook and Windows that also support older versions of Windows. Outlook 12 will provide its own improved search and will run on XP.

Along with user interface changes come end-user training issues about how to take advantage of the new search and organisation capabilities.

For example, Microsoft is using metadata for indexing and searching documents. Users will have to be educated about the use of metadata, but the initial beta releases of Windows Vista do not include mechanisms for the automated scrubbing of metadata as information leaves the machine, raising issues about user privacy.

Finally, Microsoft's search paradigm is primarily desktop-centric, whereas the need to quickly locate information spans all desktops and servers across an enterprise in a unified way.

The ability to have users run with accounts that are less empowered than "administrator" has been possible since Windows NT. However, most Windows applications came from the Windows 9x world, in which all users had full access to their machines.

Even today, Gartner estimates that 80% of Windows 2000 and XP desktops are deployed with their users running with administrative privileges for application compatibility.

In such circumstances, malicious code that the user encounters has full access to the machine. User account protection modifies the user's access token and uses heuristics to determine whether the application requires administrative access, even if the user is running as an administrator.

Vista will maintain application compatibility using techniques such as virtual registry and file redirection for users running with "standard user" privileges.

However, user account protection applies to interactive processes only, and elevates the entire process for its life, not just the period of time that administrative-level access is needed.

Microsoft's redirection techniques are good, but they will not ensure that every application will be compatible when run in a lower privileged mode, so testing will still be required. The long-term solution is to require application developers to ensure that their applications do not require administrative access and to have users run as standard users.

However, by the end of 2008 only half of enterprise Windows applications will have made this transition and standard users cannot install new software or browser helper objects, so many of today's political obstacles to user lockdown will still apply.

While Windows has had a multi-language user interface edition that companies have been buying or writing into their enterprise agreements since Windows 2000, Windows has not been language-agnostic (it still used English as its base, and not everything was translated into local languages).

Vista will use a language-neutral binary base and will treat English as a language pack. The image management tools included with Vista will be powerful, enabling image consolidation through a large degree of hardware neutrality and enabling the offline patching of images.

While some users cannot centralise image creation and maintenance because of organisational or cultural issues, multi-language user interface has already allowed some organisations to centralise images and reduce them by 90% with Windows XP. Vista's enhancements will make multi-language user interface acceptable for more organisations.

But be warned: fully exploiting  multi-language user interface may cost you more money. Specific language components may be purchased individually, or unlimited language packs are available at "no cost" with the Enterprise Edition, which will be available only to subscribers to Software Assurance on the Windows client.

Although Microsoft states that a single image could be used across different hardware platforms, drivers will still need to be managed, and it is unclear how disparate disc interfaces will be supported.

Applications that make extensive registry and file system changes and that are typically deployed using software distribution packages (including Windows Installer) will not be able to be installed to an offline image.

Neil Macdonald is an analyst and Michael Silver is research vice-president at Gartner

Read more on Operating systems software