If the service oriented architecture (SOA) was the focal point of most of the presentations and debates at the Information Age Web Services and XML Conference, then a second issue followed closely behind: What technology should, or will, be used to integrate all the various applications, components and services together?
This is by no means the only big technical issue facing those planning to implement an SOA, but it is a core one, and for many, largely unresolved. The rich and widely supported set of standards that go under the ‘web services’ banner make the task of bringing applications and services together far easier than in the past – but if businesses are to flexibly build integrated, end-to-end processes, then a flexible, reliable, secure platform is needed to orchestrate all the component services.
There are many candidate technologies that can be used to play this key role. They include integration hubs, application servers, application platform suites (such as IBM’s WebSphere, or Microsoft’s .NET), applications with integration hubs (such as SAP’s NetWeaver), web services orchestration tools and, finally, the enterprise service bus (ESB).
This last category has attracted enormous attention since its introduction back in 2001, and the importance of the ESB was the key message of two of the world’s top experts on the ESB presenting at the XML and Web Services Conference: Gordon Van Huizen, the CEO of Sonic Software, the company that invented the ESB; and Dave Chappell, author of The Enterprise Service Bus, and the chief evangelist at Sonic Software.
Chappell pointed out that many leading analysts are now backing the ESB concept. Gartner research fellow Roy Schulte, a specialist in middleware, is one: “A new form of ESB infrastructure – combining message-oriented middleware, web services, transformation and routing intelligence -will be running in the majority of enterprises by 2005 said Schulte in a much-quoted report. “These high-function, low-cost ESBs are well suited to be the backbone for service-oriented architectures and the enterprise nervous system.”
Other analysts, including IDC and Forrester, have similarly forecast that the ESB is set to play a key role in the service-oriented, web services future.
What is an ESB and why is it so important? Chappell explained that the ESB is a native technology that was developed with the SOA in mind – in contrast to other technologies, which vendors have attempted to modify in order to support the SOA.
The ESB includes several innovations: it primarily and natively uses web services standards for interlinking applications, which all treat applications as services; its architecture is distributed (hence the term bus) – there is no central integrating hub; and, crucially, it even treats the integration functions themselves, such as transformation or mapping, as services.
This last point is key, since it means customers buy and use only those integration functions they need – they do not need to buy integrated stacks of functions. This lowers the cost of ownership. An intelligent function, known as content-based routing, means that messages are read and sent to other applications, or sent for transformation and integration, without passing through a central hub.
Several delegates at the conference spoke in praise of the ESB design, which effectively mirrors many of the concepts of an SOA. Certainly, Chappell of Sonic argues that “The ESB is a pre-built SOA infrastructure with an enterprise grade capability.”
Chappell concluded: “ESB dramatically changes the architecture of integration, so you can pervasively, yet incrementally, interconnect enterprise applications through standards-based infrastructure.”
In spite of the fact that Sonic has an early lead in this market, Chappell pointed out that the ESB concepts are now so widely accepted that most companies in the middleware and integration market have adopted the concepts – including giants such as IBM. Many of these are re-engineering their products around ESB principles.