In Las Vegas for Microsoft Management Summit 2003, Microsoft Senior Vice President Brian Valentine revealed that plans for the next two Windows releases--code-named Longhorn and Blackcomb--are very much in flux. Noting that the company had "boldly" committed to shipping Longhorn as a client-only release and Blackcomb as a server-only release, Valentine said that Microsoft is reevaluating this plan and will likely ship both client and server components of each release. However, this statement doesn't necessarily mean that Microsoft is planning a Longhorn server release, per se. Rather, the company will probably ship some sort of Longhorn add-on pack for Windows Server 2003.
"There were some bold statements made at certain points about client-only, server-only," Valentine said. "But when you think about it, anytime you do a new client release and that release is either developer/ISV-focused or rich-scenario-focused (around more common storage and those kinds of things), you need server support if you are going to get \[that release\] deployed in the enterprise." Longhorn was supposed to include a new Microsoft SQL Server-based file system that would allegedly help users find data and important documents. But one question about this plan had gone unanswered: Why would Microsoft first implement this crucial technology on the client? Valentine's comments this week suggest the company has been struggling with this very question.
The problem is timing. Although desktop users expect regular Windows client releases, server customers have requested that the software giant better pace its server releases; the company responded by delaying the Windows 2003 release well beyond its original time line. Going forward, Valentine said, Microsoft had planned to skip the Longhorn server release because Longhorn is due in early 2005, less than 2 years after Windows 2003 ships.
"There could be a Longhorn server or something that you lay on top of Windows 2003 that looks like a great server for the Longhorn desktop," Valentine said. "There are options in front of us now. Do nothing: I don't think that is actually possible to get to the scenario that we want to achieve. Do service pack-level stuff: But that breaks the rules of service packs, which are supposed to be \[primarily\] bug fixes. Ship add-on overlays on top of the server kernel: I don't just mean the memory manager and stuff like that, but Active Directory and the core-level server services. Or \[we could\] develop a brand-new release of the server. Overlays sound pretty good to me in the timeframe that we are talking about, but we are not set on our plan yet."
So although the plans to skip a Longhorn server release haven't yet changed, if Longhorn slips further--a distinct possibility--Microsoft might very well add a Longhorn server release to the time line. "We'll do whatever makes the most sense and is easiest on our customers," Valentine said. "\[The plans are\] in flux. There is a wave of very interesting things that we are doing in the Longhorn timeframe and we have yet to decide the delivery vehicle for those. The goal is to be nonintrusive, or least nondisruptive, to the \[enterprise\] customer environment."
The Blackcomb Windows release, however, will include both server and client releases. Although Microsoft says it's too early to begin discussing a possible Blackcomb release schedule, Valentine discussed 2- to 3-year enterprise release cycles, which places the Blackcomb release in the 2007-2008 time frame. Given the constant Longhorn delays, however, that goal is obviously a moving target. "You have to prove the value in the infrastructure through the end-user scenarios," Valentine said. "The rate of server and services adoption without some value on the end-user devices--not just PCs but \[also\] PDAs, Smartphones, and any device you can name these days--is really slow because customers don't see the value in it."
Valentine provided other interesting details about the upcoming Windows releases. In addition to new integrated-storage technology, Longhorn will include rich new collaboration tools. Longhorn uses the same core kernel that Windows 2003 uses, but Blackcomb will probably require a new server kernel (necessitating both server and client releases). Blackcomb will also include dramatically reworked core services, such as Active Directory (AD) and public key infrastructure (PKI), Valentine said.