Embracing change never comes easily. Particularly when a jungle of post-Windows 3.1 client hardware choices is beginning to give customers a bewildering number of options.
In carving out a migration path from the estimated 81 million legacy desktops running Windows 3.1, corporate IS departments everywhere are faced with an unprecedented amount of choice, including NCs, NetPCs, Windows-based Terminals (WBTs), and a variety of Microsoft operating systems.
In some cases, the expectation of ditching 16-bit Windows and automatically upgrading to the next 32-bit Windows OS and application suite from Microsoft is already giving way to the bevy of thin-client offerings that might pave the road for a possible fundamental shift in enterprise computing.
"[Windows] NT is more expensive to run. The server is costlier, and the people you need to run it are more expensive," says Beverly Russell, an IT manager at E.D. Smith, who intends to replace her fixed-function PCs with NCs. "I don't see [IBM's] Network Stations NC replacing the PC, but I do see them replacing 486 machines running Windows 3.1."
These types of options are also giving IS managers an opportunity, albeit one fraught with pitfalls, to re-examine corporate network strategies.
"Anytime you go through a transition, you want to evaluate the overall environment," says Bruce Stephen, an analyst at IDC. "Users are no longer happy with moving to the more expensive technology."
Even Microsoft concedes that IS managers no longer are necessarily hooked into Microsoft's 32-bit Windows upgrade path.
An unappealling prospect
"Anything that comes out that could actually blow a hole in the Windows boat, those are big threats," says Steve Ballmer, Microsoft's executive vice president. "If there is a thing that can bring Windows down, it's some combination of Internet coolness, middleware OS-ish, and new device-ish - and that's what you sort of get when you put browsers, Java-class libraries, and NCs together."
Microsoft's answer to this threat is WBTs, which are on the horizon with NT 5.0. Although the specifications have not yet been released by Microsoft, the ultra-thin clients are expected to use the Windows CE operating system, a new multi-user version of NT 5.0 code-named Hydra, and the T-share protocol.
Most IS managers are reluctant to replace existing systems at all, but with support for Windows 3.1 tapering off and threatening to cease altogether in the next two years, something needs to be done.
And upgrading to the latest Microsoft OS can be an unappealing prospect for end users who don't need the robustness of high-end Pentium processors and Windows NT. NCs are an option, and as Java matures as an operating environment, analysts expect the choices to improve. But Java still isn't for everyone and Windows has proven a hard habit to break.
For example, in the absence of Windows NT 5.0, Microsoft is enticing customers with Citrix Windows emulation software on NT 4.0 servers as an alternative strategy for running Windows.
"We turned our 486 machines into Windows terminals using Citrix's WinFrame in lieu of upgrading the hardware," says Matt Gafke, a computer operations manager. "It's incredibly fast and we obviously saved a ton of money. And people are familiar with the interface so we didn't have to retrain anyone."
But the Citrix solution, which is essentially running a multi-user version of Windows NT on a server, can be costly in other ways, according to some end users. And Citrix's complicated and tenuous relationship with Microsoft is holding back the latest versions of WinFrame, according to one beta tester.
But if the 486s must be put out to pasture, what other options are viable given the fact that Windows terminals and NetPCs are a year away?
"We are saying that if you are in an enterprise that is going after departmental or corporate data and also needs to work with corporate intranets, the Internet, and extranets, we think a very viable solution is the NC," says Howie Hunger, director of channels and marketing at IBM's Network Computer Division. "A NetPC is an extension of a PC, not a new class of device. It is a PC with duct tape over the floppy drive. It is a PC with a set of packaging that limits some of the capabilities. It is not going to push NCs into a niche."
But although IBM plans to continue aggressively pushing its network-centric strategy, Hunger admits that its various thin clients are not a panacea for all corporate users. He readily admits that there are many applications where full-blown PCs and traditional portable systems make more sense, such as those involving high-end graphics.
And jumping for Java and the NC require some serious restructuring of the IT environment, a daunting prospect for any IS manager. So many companies are sticking with Redmond.
But Windows NT will clearly require more horsepower than a 486 has to offer, and that means buying more-expensive PCs with the latest Intel processor and loads of memory and disk space.
"There are a number of opportunities and changes right now," says Michael Takemura, product manager at Compaq.
However, not everyone in the PC community thinks NetPCs are viable.
Nevertheless, a number of IS shops queried said that they are already in the process of upgrading their 486 systems and don't see shifting gears to Windows terminals, NCs, or NetPCs as an immediate option.
Technical services manager Frank Petersmark says during the past 12 to 18 months his company has invested heavily in upgrading.
Although users are primarily going through a Windows migration in order to get access to more powerful 32-bit applications, Petersmark says he might still consider a Citrix-like thin-client solution in parts of his company.
But the process of moving off Windows 3.1 will take several years and is usually done piecemeal.
"NCs are still green screen replacements. People would like to try them, but the whole picture isn't in place yet," says Brian Murphy, an analyst at the Yankee Group. "Many will migrate to Windows 95 or NT and bite the bullet on hardware costs. It's a bitter pill to keep replacing PCs, but there is a light at the end of the tunnel. In two years, the other thin-client options will be more viable, and the NetPC will already be a bad memory."