Automatically linking disparate applications is one of the hottest trends in software. Among the host of new vendors pitching their wares in this enterprise application integration (EAI) segment is US-based Vitria Technology. Rather than focusing on low-level connectivity issues, Vitria's BusinessWare takes a model-driven approach to integration. Vitria's chief technology officer and co-founder Dale Skeen (who also founded publish- and-subscribe middleware vendor Tibco) spoke with IDG's Martin LaMonica about the current state and future direction of the EAI marketIDG: Why has this category of enterprise application integration emerged?
Skeen: I think business drivers and technical drivers are behind it. This has been a problem for a long time, but recently some of the drivers have become the growing acquisition of packaged applications. You're finding that ERP [enterprise resource planning] people want more than a single ERP package, or they have many packages together. That's one driver.
The second driver is increasing mergers and acquisitions. When two companies come together, they want to represent a single entity to their partners and their customers.
Their products and services should represent a coherent set of products and services. And to do that, they have to integrate their business systems. So that's another very important business driver.
I think the third, and perhaps the most important, is closer links with customers, suppliers, and partners. And the Web and Internet have really been a facilitator here.
Do you think standards will evolve for these products?
I expect that the Internet standard will be the ultimate driver in all of this, because people not only want to integrate the applications, but they also want to extend the applications out to the Web or to the extended user community, and that means working with Internet standards.
One thing that you're asking is will standards evolve? That becomes a little bit of a hard question to answer, in the sense that, when we look at integration, there are many different levels in which you have to integrate. You have to integrate the communication layer. You obviously have to be able to connect and send messages between two systems. You have to integrate the data layer.
You obviously have to be able to communicate what a customer is from one system to another, or one order from one system to another.
Then you have to have common models of business objects and business processes. Even though there will be some standardisation there, how companies represent or implement their business processes is still a very important, competitive differentiation for them, a great source of innovation for them.
So why would someone want to buy an integration product rather than write to the APIs that ERP vendors are starting to make available?
That comes down to the definition of an integration product. Some integration products define themselves as those that just provide connectors between two systems so they can translate the format and their protocols. These are what I would call connector-oriented integration packages. Over time, this cat-egory of products will just simply go away. Their pricing isn't that great, and eventually their value added isn't that great, because at the connection level, standards are going to play the major part. XML [Extensible Markup Language] is poised to take off and be very important there.
I also think that data level integration tools are also going to be severely impacted, because up until now, most integration tools have either been connector-oriented or database-oriented. That means they really are about replicating data from one database to another. Vendors are going to provide tools for that, for your standard databases, like data warehousing.
And what we're seeing more importantly is that when people talk about application integration, what they really want to focus on is how I integrate my business processes across standard applications.
Most enterprise applications are customised. Because most application integration products rely on generic connectors between applications, will they work effectively?
Customisation needs to be accommodated. And the challenge to the EAI product is to make that accommodation easy. And that is really the vision we want to work toward - once the plumbing problems go away, the higher level problems become more evident at that point in time because they become much more visible.
What kind of impact will EAI products have a few years from now?
I think this is the next big thing. This is as important as the introduction of databases were in the '80s, when people used them to automate business tasks and departmental systems.
This is as important as the introduction of the ERP systems in the '90s. EAI will be the next big trend and will be the dominant technology force in the late '90s and early in the 21st century.