In case you missed it, IBM's Software Group is very big on applications built using Web services and adhering to a services oriented architecture (SOA). And, as the largest supplier of legacy applications running on mainframe and i5/OS and OS/400 platforms, and a big supplier of Windows and Unix platforms (which also have their own legacy issues when it comes to interfacing with the Web), this all stands to reason. IBM's SOA strategy is as simple as it is predictable--add another layer of abstraction atop of legacy platforms, exposing their fields and data streams as services, and then allow companies to weave them together in new ways.

The brilliance of this approach from a business perspective--meaning IBM's business as well as your own--is not to be denied. By moving to an SOA approach, you are basically preserving application platforms as they currently are, which gives those legacy applications some reprieve--at least for now--from the pressure to upgrade them or dump them for something else. Rather than rip and replace, which is expensive, time-consuming, and of dubious value because massive ERP and related application projects fail as much as they succeed in the real world, IT shops of all sizes are being asked to try out this new way of building applications and not tear their shops and applications apart. They are mapping their corporate application genome, and recombining the genes in new and interesting ways, to use an analogy. And they are moving up yet another level of abstraction off the underlying iron. This does, of course, come at a cost. Something has to run that abstraction layer, and that is, of course, more IBM server horsepower. And that leads to more database, middleware, and development tool sales.