By Anne Chen
When Steve Broudy began introducing Windows 2000 Professional into his company's desktop computing environment last year, he knew the enhancements in the new operating system, such as better security and stability, would make an accelerated migration from Windows NT worthwhile.
Today, two months after Microsoft Corp.'s launch of Windows XP, however, the upgrade strategy for the CIO at Mann Theatres Inc., in Encino, CA, isn't nearly as clear. Microsoft has also announced plans for two succeeding operating systems, code-named Longhorn and Blackcomb, to be launched tentatively in 2003 and 2005, respectively. At the same time, IS managers must consider new Microsoft licensing terms that reward more frequent upgrades with lower license charges and the fact that the software company will drop support for Windows 2000 in 2004, leaving Broudy little time to test Longhorn before upgrading, should he decide to skip XP.
After taking its sweet time rolling out Windows 2000 last year, Microsoft has returned to a pattern of releasing a new desktop operating system every 18 months or so. That pace, new licensing terms from Microsoft, and a heightened need for most IT organizations to cost-justify all infrastructure investments have combined to make it difficult for CIOs such as Broudy to come up with a long-term desktop migration strategy that makes sense. So, rather than selecting a strategic desktop operating system platform and attempting a wholesale migration to it, many IT managers are accepting the fact that, for the foreseeable future, they'll be managing multiple Microsoft desktop operating systems at once while playing a tactical game of deferring upgrades as long as possible.
Experts say that approach makes sense for most enterprises, even though having multiple operating systems usually means higher IT support, training, testing, and configuration costs. Gartner Inc., in fact, estimates that running two desktop operating systems costs on average 5 to 7 percent more than the cost of running one.
In many organizations, however, those costs can be offset by deferring operating system upgrades. It costs $230 to $500 or more for labor and licensing to upgrade a desktop from Windows 9x to Windows 2000, Gartner estimates. And that doesn't count desktop hardware upgrades required to run new operating systems. Many organizations learned, after leaping from 9x to Windows 2000, that associated hardware costs can be considerable. They found themselves saddled with labor and hardware costs when they realized existing hardware would not run the software efficiently.
Where to start?
So what factors should IT managers consider as they mull the tactical decisions they'll need to make between here and Blackcomb? The first, experts say, is how far along you are in deploying Windows 2000. The further along you are, the less you need XP and the more sense it makes to wait for Longhorn, even though Microsoft has plans to discontinue Windows 2000 support very close to the time at which it has tentatively scheduled the rollout of Longhorn.
"Our advice is, as much as possible, try to get the latest operating system you can," said Michael Silver, an analyst with Gartner, in Stamford, CT. "The average life of PC hardware is being extended, and we have a lot of enterprises with three-year upgrade cycles now starting to look at four-year upgrade cycles. Upgrade strategies should reflect that."
That advice holds, Silver said, even though Microsoft will continue releasing a new operating system about every 18 months and support each version for about four years.
"Microsoft has shipped commercial operating systems each year for four years running now, but enterprises have shown an inability to implement and a dislike for managing too many operating systems," Silver said. "It's just not feasible to begin a two-year upgrade and expect all desktops to be current with Microsoft's release schedule."
For its part, Microsoft contends it is not releasing operating systems faster than it traditionally has or faster than most enterprises need them. At the same time, however, Microsoft officially acknowledges it is not feasible for most enterprises to upgrade in a wholesale fashion to every new operating system. Officials at the Redmond, WA, software company said its release schedules are based on customer requests for enhanced features and technologies to be included in new releases of desktop operating systems.
Charmaine Gravning, XP product manager, said Microsoft views this latest operating system as an opportunity for IT managers still running Windows 95, Windows 98, and older versions of NT to upgrade and skip Windows 2000 entirely. At the same time, Microsoft also recommends that companies that have already invested time and effort testing and deploying Windows 2000 continue with that upgrade pattern, delaying any move to XP.
But for those who decide to play a waiting game, it's not clear exactly what they'll be waiting for. Microsoft hasn't discussed what licensing terms might apply to Longhorn and Blackcomb, nor has the company made any announcement regarding what new features the operating systems will have.
Gartner's Silver said he regards Longhorn as a minor upgrade from XP. Blackcomb, on the other hand, despite Microsoft's lack of detail, is being called by many the next major Windows release. As eWeek has reported, Blackcomb may include a new user interface, more .NET integration and a new file system.
Though details are sketchy, these reported new features are enough to persuade IT managers, such as Nelson Ramos, to minimize migration to XP and hold out for following desktop operating systems, assuming, of course, that Blackcomb receives support from third-party software developers.
Ramos, vice president and CIO at Memorial Hospitals Association in Modesto, CA, and an eWeek Corporate Partner, has approximately 80 percent of his 900 desktops and laptops on Windows 2000. He does not see enough enhancement in XP to justify a wide-scale upgrade.
While he has tentative plans to eventually upgrade the 50 or so laptops in his organization to XP, Ramos said the strategy in place now is to skip XP entirely and wait for Longhorn when it comes to his desktops.
Longhorn, which he views as a routine system upgrade, will be the steppingstone to an eventual upgrade to Blackcomb, Ramos said.
"I see XP more as a significant advantage for the consumer because it's solid and has a lot of multimedia that makes it very attractive," Ramos said. "But in a corporate environment, I'm looking for reliability, networking, and security, and I've got all that built into Windows 2000."
While Silver at Gartner recommended that organizations already well along with their Windows 2000 migrations continue along that path, he warned that there are dangers in skipping XP.
Silver said Microsoft's plan to drop support of Windows 2000 in the first quarter of 2004 means that enterprises that plan to skip XP and jump on Longhorn may have only a few months to test it before Windows 2000 support dries up. In fact, organizations just starting a Windows 2000 upgrade now will not finish early enough before needing to upgrade again to ensure technical support, Silver warned.
These are the concerns that plague Broudy at Mann Theatres. They're why he hasn't yet decided whether to upgrade broadly to XP or wait. Broudy's current desktop computing environment is a mix of NT 4.0, Windows 2000, and XP. Each time a new desktop operating system has been introduced, it's entered Broudy's organization via new hardware purchases.
Broudy guesses that Microsoft's decision to drop support of NT will eventually put tremendous pressure on him to dedicate limited time and resources to upgrading all his desktops. "I would rather not upgrade right now, but Microsoft keeps moving forward," he said. "We don't really have the resources to upgrade right now, but at the same time, we can't afford to be left behind."
A fear of falling too far behind the curve is one reason Frank Calabrese, manager of desktop strategic planning for Bose Corp., in Framingham, MA, will eventually bring XP into his environment on newly purchased desktop machines.
Calabrese, an eWeek Corporate Partner, has 54 percent of his 2,600 users in North America on Windows 2000. The remaining users are on NT 4.0 and a combination of 9x.
While Calabrese is testing XP and has plans to gradually phase in the operating system after the middle of next year, he has no plans to standardize on the new operating system. The reason: It is more expensive to upgrade to one software platform than it is to support two different versions.
"Our belief is that interim upgrades and paced, planned migrations are the way to go," Calabrese said.
"I want two operating systems to overlap at any given point because I can't land on a single product," Calabrese said. "If all I was doing was reducing support costs, I'd land on one operating system. But I need to provide the right set of tools for my user environment, and two operating systems allow me to efficiently manage the cost of a migration."
Even Microsoft's more deeply discounted licensing programs, such as its Enterprise Agreements and its proposed Software Assurance plan—which tend to encourage rapid operating system upgrades by users—are not enough to persuade many IT managers to jump on XP before Longhorn and Blackcomb come out.
Memorial Hospitals' Ramos, for example, is wavering because signing an Enterprise Agreement with Microsoft would mean purchasing XP licenses and upgrading all his desktops.
Remaining on his Microsoft Select contract, on the other hand, would allow his organization to upgrade at its own pace, although operating system licenses would be more expensive.
"We're trying to balance the cost of supporting a user on an older operating system until the hardware is replaced vs. upgrading and doing memory upgrades to support the software," Ramos said. "The latter seems more expensive."
Experts say that while IT managers would be wise to keep an eye on Microsoft's release schedule, they should be the ones setting the pace of their upgrades.
And, although he does not recommend that organizations get too far behind, Silver said there's nothing wrong with letting older operating system versions remain in place until organizations are ready to replace them.
"Consider how aggressive you are in adopting new technology and upgrading to new applications," Silver said.
"Microsoft's grand vision for everything is in Blackcomb, and that's not coming out until 2005. In the meantime, if the majority of your users don't need the newest applications, then proceed with business as usual."
ZDNet TechUpdate originally published this article on Dec. 5, 2001.