Cloud Computing and.. mainframes!

Do we really need to extend cloud computing to every device?

In “Cloud Computing Comes to a Desktop Near You”, I thought aloud about the direction that Microsoft, Google, and others are trying to take us. Do we really want or need to cloud-itize (cloudify? to make clowdsy?) every software process? Well, probably, yes. If we can securely and cost-effectively distribute IT to more efficient handlers of that IT (e.g., Microsoft or Google handling email and collaboration, Amazon handling storage), then there are some very powerful financial motivators at play.

But how do we get from here to there?

The move to the cloud is going to take a long time, and there will be many steps along the way. And, for some business processes and functions, we have to realize that we’ll keep some things in-house, even while we take advantage of what web services and cloud computing is bringing us. (A more full-featured, available-to-all middleware infrastructure.)

Let’s take GT Software’s Ivory Service Architect as an example. GT has developed software that is supposed to make it easier to take an existing business function that is available to your application and turn it into a web service. Your software can run on Windows, Linux, or even IBM’s System z mainframe family, all while being able to access existing business services on your mainframe system, which you’ve probably spent a lot of time and money on in terms of application development and tuning.

Mainframes? Web services? But why?

Mainframes are still running a lot more than you think, and they’re still being actively deployed. For example, if you look at large hospital systems, insurance firms, state, and Federal government agencies, you’ll find mainframes.

Really though, aren’t we just exporting new API’s to old services?

Well, yes.

And that’s part of the migration strategy. Why require that your application run on a mainframe to access a service provided by another application running on a mainframe? It makes no sense. Instead, provide a layer of abstraction to access that service and let that be enough.

Again, this isn’t a new concept. We’ve had API’s that are network-accessible since, well, the beginning. But clearly the focus is now on moving everything toward that model, not just targeted subsets of your software infrastructure.

Interesting things to come.

• Microsoft
• cloud
• future
• mainframe
• Web services