A Focus On Frameworks Makes Platform Choices Less Odious For Developers


Linux and open-source components pave the way


In the next few years, resellers and ISVs will have some of the best platform choices ever. Choosing the OSes and software stacks to deploy their applications has generally been a tricky--and risky--business, but recent events and market trends offer a more sanguine set of alternatives.

A large part of this elevation of choice has been the surging effect that Linux and open-source components have had in the market. Another has been a wider use of virtual-machine-container technologies such as Java and the Microsoft Common Language Runtime. Yet another has been the embrace of Web services specifications and a growing appreciation of the long-term benefits of a services-oriented architecture (SOA) approach. The common-architecture espoused by grid, on-demand and utility-computing proponents are helping, too.

The ability to create and customize applications and services with less intrinsic interdependence on a specific OS has, in fact, forced a new form of philosophical choice for developers and architects: Do I gain more for my applications and ability to make money from a highly evolved OS that includes more integrated capabilities? Or do I gain more by hitching my star to higher-level frameworks and loosely related but accessible components that adhere to standards or widely embraced specifications?

The best answer is both. However, whichever one ISVs and systems developers settle on has the potential to reshape the IT software business. ISVs are eyeing enterprise Java on Linux as a hedge to platform diversity, while looking to Web services to allow interoperability of content, data and logic across frameworks.

Also at work in the marketplace has been the recognition from many server-components and business-applications vendors that they can commoditize the platforms beneath them via adoption of Linux. These ISVs keep more money from their customers while creating application-platform combinations and adding on professional services revenues. The perceived value remains with the applications. Microsoft's plans for Longhorn, the next generation of Windows, has done little to keep ISVs from resisting Linux and platform commoditization, according to Yankee Group research.

More important than Longhorn is the emerging grease on the skids to SOA benefits. For example, some companies are showing how legacy Cobol and CICS code can be deployed with great performance and fractional platform costs on Linux and x86 blades. Use the code; drop the mainframe. It's the best of the old and new.

Budding platform-plus providers are showing that managing and manipulating the abstraction layer above the platform and the applications is where the new value-add is. Keep applications but reuse them and combine their assets in new ways. Even IBM, the Lothario of legacy, is into the "abstraction above" philosophy, as its recent announcements on content/data (Marsala), SOA, virtualization and modeling tools attest.

Together, these examples demonstrate the potential for the next stage in IT productivity: Take existing application assets and deploy them on low-cost platforms or grids, then service-enable and reconstitute the functional attributes for a higher level of business-productivity benefit. Will OSes be able to accomplish this?

The recent April 2 bear hug between Sun and Microsoft may itself be a hedge on how this philosophical shift plays out. New, closer ties between the companies could diminish the IBM-burnished luster around Linux. Such Linux momentum has clearly helped push Sun and Microsoft into each other's arms. Why not allow Sun Solaris on x86 to play well with Windows on x86 platforms, providing a richer cross-platform alternative to Linux?

But the connection between Sun and Microsoft could, and should, accomplish much more than just trip up Linux, which is a symptom more than a cause. That's because of the power inherent in building and orchestrating applications and services--new and old--regardless of any specific OS. Movement is not just to Linux; it is to a richer set of architectural alternatives. Consequently, there may be no turning back to the days of platform-tool-application lock-in.

The upside is that the relevancy of the monolithic operating system is in flux. And a shift away from such do-all platforms does not necessarily mean leaving the application code and intellectual assets behind. It means a better set of choices for ISVs, integrators and enterprise architects.