The Connected Web

Traditional IT management systems take an approach that won't work when it comes to cloud implementations, says Dan Foody, VP of products for Actional at Progress Software. Historically, management has started with the underlying infrastructure and worked upwards through the computing stack. "When you move to the cloud, that breaks down significantly," he explained when we met recently in London. Cloud providers don't want to expose the lower echelons of their infrastructure, and in an architecture where service contracts define the interfaces between components, customers shouldn't have to care what's happening lower down, so long as the service contract is being fulfilled.

When dealing with cloud services, he said, "customers don't worry if their IT systems go down, so long as their business transactions continue to work." Think of the servers in a cloud data center as trees in a forest. When one fails, it's a silent, unnoticed event to the outside world. Another one just takes its place.

The problem is that this philosophy is alien to the way most IT management systems work. They want to be able to see the entire infrastructure. ITIL, which defines a framework for IT service management, doesn't provide a solution because it, too, is looking at the infrastructure from the bottom up. "ITIL doesn't go far enough in many ways," said Foody. "It stops at the IT layer and doesn't really look at the transaction layer."

What Foody believes organizations must learn to do is to reorient the way they think about application management to focus on the transaction layer as distinct from the underlying IT infrastructure. "People need to be conscious of and understand their business transactions end to end," he said. "If you're running across Google Apps and Salesforce.com, who's going to worry about the end-to-end business transactions?" The providers of the underlying infrastructure aren't going to do that for you.

Actually, he says, if you stop using the IT layer as a starting point, it gets a lot easier to solve many problems. By focusing on the business layer rather trying to solve things from the IT foundation upwards, Foody said that one Actional customer has cut the headcount needed to track operations from 20 people down to three, as well as halving the time spent by those three to resolve each incident.

Foody was in town to talk about the release of Actional Diagnostics, which is an upgraded and rebranded version of MindReef's Soapscope testing tool, acquired by Progress last year. One new feature that interested me was the addition of 'Application X-Ray', some Actional technology that allows people to see the 'knock-on' effect of a service request. "That gives people this visibility into what happens to the application once they send a service request into it ... Rather than your application being a black box, it's a white box, so you can actually see what's happening." This is much better than traditional debugging or logging for testing composite applications, he explained. It's especially useful when, for example, calling a service like Google Maps and then perhaps another service as part of a composite process mashup. Of course it doesn't tell you what happens inside Google, but it allows you to see what you send in, and what comes out.

Actional Diagnostics also adds REST and POX to Soapscope, a telling indication of these interfaces gaining in respectability. Enterprise developers often prefer them when linking to third party resources or when building the UI layer, Foody said.