A BREIN for business - an interview with John Brooke

Yesterday I spoke to John Brooke from the University of Manchester, who works on the BREIN (Business objective driven Reliable and Intelligent grids for real busiNess) project. BREIN's aim is to link business processes to grid computing. By taking the concept of "dynamic virtual organisations" and enhancing the system with methods from artificial intelligence, intelligent systems and the semantic web, BREIN hopes to enable businesses to easily and effectively use grid technologies.

The BREIN technology is currently being tested in two specific scenarios – in engineering and airports. In the first scenario ANSYS, a provider of engineering software, is working towards selling its software as a service rather than a product using the BREIN technology. The second scenario involves BREIN modelling processes in Stuttgart Airport (such as pilot timetables and shuttle bus scheduling) in an effort to reassign resources if, for example, a flight is delayed.

In a session yesterday at the eChallenges conference John explained how both grids and clouds were helping ANSYS and Stuttgart Airport. To make grids more accessible for business, BREIN is trying to bridge the paradigm between grids (sharing resources across boundaries) and clouds (virtualisation of resources). Their objective is to provide a model of switching between the resources, so that clouds can be treated as grids and a company such as ANSYS could easily outsource jobs to places such as Amazon rather than having to purchase more of its own computing power.

After the session John told GridCast a little more about BREIN and its work with grids, clouds and business..

GridCast: Do grids have a lot to offer business?

John: Yes I think so. The point about this project is that they've got a lot to offer business if they fit in with the business process. For example if you're paying for computing you want that model of payment to go on the normal accounting mechanism so that you don't have to develop a special means to pay, for example, Amazon. As far as [a business] is concerned it's like a normal cost budget item.

What we have in BREIN is a business logic, which is a high level logic – what the business wants to do. We then transfer that to the technical language - which is how do you do it. The fact that we've got this connection between the two means that if the business plan changes, the technical plan changes. So to give you the airport example, the business plan is that the airplane will land, the baggage will arrive, the tickets will be validated and, if you're hiring a bus, the airport will pay the bus company and will reclaim it from airport fees, etc. That's the business logic – it doesn't change if the plane's late. But if the plane is late the buses you actually get change, the times you get picked up change, so it's separating the business logic from the actual logic of implementation.

If we can get it right then we should be able to describe the business process in terms of our business ontology. And we have a way of mapping that down to the technical requirements. One of the big problems with grid is that it's all a bit technical so a company can't see how it relates to the normal way they evaluate a project, which is how much money they're spending and what's the benefit for their business. We think we're allowing them to integrate the use of computation in such a way that they can make that decision and all the technicalities are hidden. And that's a research aim because it's not completely established that that can be done.

GC: Where do clouds come into this?

J: Basically a cloud is a very good way of getting hold of resource cheaply and at short notice. So in the case of our engineering application probably the engineering firm doesn't want to buy a computer, possibly ANSYS doesn't want to buy a computer, but somebody like Amazon or Google have huge banks of computers. And Amazon's computers are heavily loaded at different times. If you think of the classic grid like EGEE, it's made by joining lots of clusters. So if you replaced one of those clusters with the Amazon cloud you've got the idea.

GC: What are the main challenges for this?

J: The main challenges are that people describe things in different ways. So let's say you've got an urgent need to run something by a particular date. One resource provider might give you something like 'termination by' while another might give you 'completion by'. If you compare those statements to find out if they're implicitly the same the answer is no they're not. But they clearly mean the same thing. So what we're trying to provide is the semantic technology where we can take a process, implement this in some ontology language and that can then be converted into things like xml files. You can then take those two statements, check them against your model and say look these are the same thing. So you can then switch from one provider using one language to another provider using another set of terms.

BREIN works on doing this dynamically. In the airport scenario we allow reasoning – we allow the plans to change. Let us suppose you've got a plan for a plane landing. Everything's worked out, you're going to get the buses on time, then you find out its delayed. You've got to re-plan. Not only that but you've got to do it dynamically because you don't have forever to re-plan, you've got to do it within a time constraint . If it's too late after the plane's landed then its no use getting results. You have to be able to dynamically reprovision all these things.

Thanks for talking to us John! For more information about the BREIN project, including a rather snazzy and illuminating video visit their site here.