Informix Founder Sees Composite Application Bonanza

It's commonly believed in the software industry that composite applications will become the primary means by which corporate users will interact with a company's business processes. That vision drew serial entrepreneur Roger Sippl out of self-imposed exile in the venture capital community. As a founder of companies such as Informix, Visigenic and Vantive, Sippl has extensive experience with data access and management applications. Today he is CEO of Above All Software, a provider of tools for creating composite applications. In an interview with CRN Editor in Chief Michael Vizard, Sippl talks about the trends he says will profoundly reshape applications as we now know them.

CRN: What brought you back to an active role in the industry, founding Above All Software?

SIPPL: I read about the Web services standards and about how all of the business logic that people now build in application servers will be available as standard-looking software robots that you can send a message to. So I thought: That means you can give it an input, like a customer number, and it'll give you back the customer's name, address, phone number and everything else about the customer. And maybe there's another Web service that'll run a different application server. You give it a customer number, and it gives you back all of the information about the customer's orders. And then you take the order numbers out of those orders, and you send them to a different server that can tell you whether those orders have shipped. And I just had an image of those different servers needing to be composed together with some easy-to-use tools for composite views of a transactional enterprise. I realized there's an opportunity here that's every bit as big as the relational database management system opportunity in 1980.

CRN: Do Web services make that possible?

SIPPL: The Web services standard makes it easier. It doesn't solve all the problems. We solve the rest of the problem. The Web services standards make it so that there's one industry-standard way to send a message and have answers come back from a query, or have a commit come back that the update has succeeded. It gives us one common language, but the models are still different between different silos, such as Siebel and SAP. The Web services standard takes care of some of the plumbing and some of the smaller details like, how is an integer formatted. You don't have to worry about the byte order of a 2-byte integer anymore because the Web services standard takes care of that. But we still have to worry about the fact that you might need to re-factor business logic. That's the missing piece in getting from the concept of a services-oriented architecture to an easy-to-use composite building system. You have to have a modeling layer in there, and that's what our product does.

CRN: What does your software do?

SIPPL: By using a nice graphical user interface product like ours, you can say, 'OK, go to this server, tell me about all the Web service operations on it,' and we catalog those into our dictionary. Then you can build solutions by just dropping those operations into forms. We automatically render solutions that let you type in a customer number and, boom, you can see the guy's customer information. And then from the order system you can interrogate its Web server and find out what operations it has. And maybe it has, 'Get orders. Back order something. Cancel an order.' You could drop new orders, cancel orders and back orders into that same form, because those are all Web services capabilities.

CRN: It sounds like you are trying to get to the point where a business systems analyst can integrate business processes without having to have someone write some code?

SIPPL: That's my Holy Grail. We're on version 3 of the product now, so I think we're most of the way there. This is the beginning of that era. We let you point and click together several Web service operations, and we have an engine that executes the composition of those operations to do a high-level task. We can create the object of a customer and his orders even though that might touch three Web services in the customer system and four Web services in the order system. That means business analysts can solve their own problems.

There's nothing stopping us from building the same thing I built at Informix. We have an engine that can relate customers to orders and shipments. Just like a relational database relates those tables, I can relate whole silos together. And just as business analysts can use a report writer to create reports off relational databases, it's just that easy to talk to these three silos and create a report or a transactional application across three silos. It's almost really the same product. It's been scary to me over the last three years how much I'm reliving Informix.

CRN: Given that capability, will we see a resurgence of best-of-breed applications, as opposed to suites, from one vendor?

SIPPL: I absolutely believe that best-of-breed is what people always wanted--and they always will. This does enable the best-of-breed opportunity to come back. Even if somebody standardizes on Oracle or SAP, that only covers 40 percent of the applications they have in their corporation. The 60 percent of the home-grown applications cannot be replaced by buying an application anyway. Those applications tend to be very specific to the functioning of that company. We'll always have the need for tools that integrate best-of-breed things. Products like ours make it easier to integrate them together. That's the power of standards like Web services.

CRN: Could that ultimately lead to a commodity software market, where one component can easily be swapped out for another?

SIPPL: I think that may start to happen in my lifetime. But it takes a long time to go from innovation to commodity. I think it will take awhile for the business logic layer of SAP or Siebel to become commoditized. In the meantime, I think customers will now look harder at the business logic layer. The evaluations used to be solely on the user interface. I think those days are numbered. Right now people look at an SAP order module, or look at Siebel CRM, and they see a lot of complexities. I think once they are able to model that complexity, they'll get a much better understanding of those models over the next 10 years.

CRN: What's your advice to companies in the channel as this trend develops?

SIPPL: I think every systems integrator should be building a composite application practice. You might call it an SOA practice or a Web services practice—but those are technologies. If I were them, I would call it their composite application practice because that's really why you would build an SOA. If I were running a systems integrator company, I would do exactly that.