I've been beating the drum for Windows Server "Longhorn" for quite some time now: This next Windows Server version features an incredible roles-based installation and management model, sweeping new functionality, heightened security, and a slew of other features that I think will make this the most rapidly adopted Microsoft server product ever. But last week at its Windows Hardware Engineering Conference (WinHEC) 2007 trade show, Microsoft took the wraps off what is arguably Longhorn's most important feature of all: its name. No, Microsoft didn't whack itself with a crazy stick and brand this release as "Windows Server Vista" or something equally inane. Instead, Longhorn will simply ship as Windows Server 2008. Apparently, maturity and normalcy won out in the Windows Server division at Microsoft, at least on the branding front.
On the flipside, the Windows Server division also used WinHEC to spread the word about another, less exciting, change to Longhorn: That hypervisor-based virtualization solution the company has been touting for a few years now will still allegedly ship within 180 days of the date Windows 2008 ships. There's just one problem: It will do so without three of its core features.
I can hear the collective groan from the audience. The problem is that there's a perception that the Windows Server team has historically under-promised and over-delivered. (Compare this with the Windows client team, which, well, let's not go there.) In truth, that perception doesn't match reality. Microsoft dropped high profile features like Network Access Protection (NAP) and various Terminal Services improvements from Windows Server 2003 R2 to meet that product's schedule, and the dropped features we're seeing now in Windows Server Virtualization are roughly analogous to that. If there's a positive spin to this at all, it's this: The Windows Server folks certainly take the regularity of their schedule very seriously. And if shipping these technologies later than expected results in higher quality, maybe this isn't such a bad thing after all.
Maybe. There's just one problem: The features being dropped are among the most eagerly awaited features. These include live migration capabilities, hot-adding of storage, networking hardware, memory, and processors, and support for up to 32 processor cores; now, Windows Server Virtualization will support just 16 processor cores. When you combine this with the news (admittedly not new) that Microsoft's upcoming virtualization management software, System Center Virtual Machine Manager 2007, won't be free to licensed Windows 2008 users, it all adds up to a deafening thud. Suddenly, Windows Server Virtualization isn't quite as exciting or full-featured as originally expected.
This is, of course, a huge opening for VMware and a sign to VMware's many customers that their technology investments are quite safe for the foreseeable future. If you've been waiting on Microsoft's hypervisor-based virtualization solution, or perhaps biding your time with Virtual Server until the Microsoft hypervisor ships, this month's bad news is at least an excuse to reconsider your options. The problem is simple: Even when Windows Server Virtualization ships in 2008, it will lack many key features that are present today in VMware's ESX Server. On the other hand, Microsoft's technical approach to virtualization, which should quickly result in better driver coverage than is possible with ESX, could ultimately prove to be the superior solution.
In case it's not obvious, there are no easy answers here. And at the end of the day, that's the biggest problem with this announcement. It makes any decision on virtualization technology even more difficult than before.