How Far Can Microsoft Spread its .NET?
Will Microsoft's .NET eventually bring mainframe and midrange data into the wonderful world of Windows?
What exactly is .NET? A consultant with a leading IT services firm I recently spoke with likened Microsoft's .NET vision to "a spell-checker for rent over the Internet for Word." Another analyst views it as a Windows-preservation strategy in a world moving to Web services—and potential platform independence. Maybe Microsoft hasn't done a good job explaining .NET, or maybe the software giant wants to keep the concept in a formless state until it sees what the market is willing to pay for.
In October, I reported on WebSphere, IBM's integration machine. One of its biggest competitors, of course, is Microsoft, offering a constellation of tools ranging from .NET to COM+ and Host Integration Server. While WebSphere fits snugly into IBM's established base of large systems or legacy environments, analysts say Microsoft isn't quite ready to take on the enterprise with .NET anytime soon. Some Microsoft partners, however, claim they're doing just that, with mainframes to boot. Can .NET move beyond Windows and become a viable infrastructure for large systems integration?
For starters, it's important to note that while Microsoft has attached the ".NET" moniker to everything it churns out, there aren't a lot of truly .NET product offerings out there beyond development tools. "Visual Studio.NET is where .NET becomes real," says Mark Driver, analyst with Gartner Inc. Its most tangible features include the new Java-like C# programming language, as well as a common language runtime (CLR) bucket that supports COBOL and other legacy code. Ultimately, .NET's Web services foundation may help bring Microsoft technology closer to all systems supporting Web services standards.
However, for the present, .NET seems to offer no particular advantage to larger (non-Microsoft) system sites. "As of today, there's nothing in .NET that makes it intrinsically better for environments with heavy legacy code than any other environment," says Driver.
"Microsoft doesn't have the deep tools that IBM has for getting to CICS or IMS data," agrees Ted Schadler, analyst with Forrester Research. "Microsoft typically relies more on partners to build adapters to that data, or relies on standards like JDBC and Web services."
This doesn't mean Microsoft will forever remain out of the large-systems, host-integration game, of course. In fact, Microsoft seems to be taking a cue from IBM's enterprise modernization strategy, Schadler says. "Web services technology is fast enough and robust enough to use for many data integration problems, data exchange and portal interfaces. That means that companies can re-purpose the HTML interfaces they've already put on their mainframes as XML, and add new XML interfaces to their mainframe data."
Host integration projects between large systems and Microsoft platforms are possible today without .NET, of course. Currently, host-integration technology available from Microsoft consists of Host Integration Server and MSMQ adapters. In addition, IBM offers its own CICS adapters for Windows operating systems. "The bottom line here is that .NET is largely still a wrapper over existing COM+ services," says Driver. "You use the same technology that you used in the past to do legacy integration, with .NET wrappers around those same services."
While analysts are skeptical, Microsoft partners seem to feel that .NET is a step in the right direction to bringing mainframe- or midrange-based data into the Windows world, especially with support for Web services standards such as SOAP and XML. "SOAP allows for application-to-application communication over standard transport mechanisms, like HTTP," comments Michael Pelletier, emerging technology strategist with Pinnacle Decision Systems. "If the ‘legacy' system supports HTTP, and there's an XML parser for the operating system we are trying to interact with, then we have a way to interoperate."
One Microsoft partner, Extreme Logic, has already integrated Windows-based applications with its four mainframes. Paul Hernacki, strategic consultant with Extreme Logic, feels that Microsoft's .NET "already is a viable approach to integration with legacy-based systems such as mainframes, midranges and high-end Unix systems. .NET helps avoid a ‘rip-and-replace' of everything an organization already has in place. It enables people to build new systems and applications that rapidly tie together existing systems with new state-of-the-art capabilities and functions."
Other partners are also applying .NET-based interfaces to large systems integration projects. "We use .NET to create a new application framework to create reusable code objects that multiple applications can use to access transactions on mainframes, midranges and Unix systems," says Paul Thomas, consultant with Crowe Chizek, an Indianapolis-based systems integration firm.
Microsoft's Achilles Heel
Nonetheless, analysts contend that .NET is too closely tied to Windows. "There's nothing in .NET that makes it more attractive to legacy integration, other than Web services, than the old COM technology," says Gartner's Driver. "Microsoft's Achilles heel, when it comes to enterprise environments, has always been that its technology runs on a single operating system family." By contrast, technologies such as Java now can run natively across all IBM and Unix platforms.
While .NET will remain closely tied to Windows in the near term, it may take on a more platform-independent form within the coming decade, Driver predicts. "It's inevitable that .NET will move beyond Windows. If Microsoft doesn't do it, someone else will, through reverse-engineering or cloning the platform."
For organizations relying on non- Microsoft systems, .NET has a way to go before it becomes part of large enterprise-integration projects. For Microsoft-centric shops, however, it's the path of least resistance to the Web services space.
Joseph McKendrick is an independent consultant and author, specializing in surveys, technology research, and white papers.