Software to integrate old data now comes under the label of service oriented architecture. The IT industry has historically produced two main responses to the integration question: more software or more ‘architectureware’.

In the first category consider such phenomena as 4GLs (proprietary languages to ‘speed up’ application provision, allegedly by non-programmers); computer aided software engineering (CASE); and object oriented (OO) languages such as C++ and later Java. Some of these approaches are now part of history, while OO is still alive and well in the shape of Java Enterprise Edition version 2 (J2EE) and Microsoft’s Csharp. In any case none of these, often excellent, development approaches was the ‘silver bullet’ solution promised.

In the second category, architectureware, enterprise application integration (EAI) and now service oriented architectures (SOA) may be placed. EAI grew up in the 1990s against the background of enterprises acquiring portfolios of systems from the likes of Baan, Peoplesoft, Oracle and SAP. The idea of linking many applications together through a common EAI platform such as from Tibco or WebMethods has proven to be of some interest to large companies. Undoubtedly these solutions are not common or garden shrink-wrapped offerings. Analyst firm Gartner has estimated the average integration product cost at around $400,000 each in 2002; many commentators caution that the ratio of licence price versus consultancy to get such systems fully operational could be in the order of a factor of eight.

"Good as EAI, SOA and ESB are, they are typically positioned by the IT community primarily as a way to link new applications, more usually web-facing ones"

From strength to strength

Nonetheless there is a strong EAI market. In 2003 IDC predicted that by 2006 the integration market would grow from 2004’s $6 billion to $8bn, a 17 per cent compound annual growth rate. The reasons, say the analyst group, include the continuing ubiquity of heterogeneous computing environments, a move toward fuller application deployment suites prompted by ongoing complexity of implementing multiple ‘classic’ point-to-point products, and “a need for industry-specific, role-based, and/or size of target organisation business process automation functionality”.

Meanwhile SOA has been gaining ground in IT circles. The idea behind SOA is that applications are less important going forward than the services they provide to the business, so an architecture or overall framework of same would be a collection of services that communicate with each other using web-based formats. Many commentators note the lineage between today’s SOA concept and its OO predecessor, Microsoft’s DCOM or Object Request Brokers working with the old Common Object Request Broker Architecture (CORBA) spec. What is important today is that use of Microsoft’s .Net framework and the open systems community’s Java J2EE competitor are typically seen as the first development steps to an SOA platform.

Note that there is no ‘competition’ between EAI and SOA or web services as the approach is also sometimes called. The best way to see web services is to view them not as a replacement for EAI or even something to solve all your integration problems, but as an extension of the integration story and a ‘plumbing’ technology that can simplify some of the more mundane tasks of an integration initiative. Some even see SOA/web services as EAI in another guise, this time for the smaller midsize enterprise.

An example of how they are being bridged, for instance, is the Java-based Enterprise Service Bus (ESB). The idea is that applications are connected to a more lightweight backbone than the traditional EAI platform, which then takes care of functions such as intelligent routing and transformation, connecting and coordinating the interaction of services across highly distributed contexts.

But good as EAI, SOA and ESB are, they are typically positioned by the IT community primarily as a way to link new applications, more usually web-facing ones. While spot on for that class of application, it may not be of too much immediate help to those users struggling to integrate big back-office systems, let alone legacy or bespoke software. Of course such work can be done ‘natively’ by in-house developed software bridges or adaptors, but the danger with this approach is that it necessitates internal resource and is by definition a somewhat ad-hoc and non-standard solution that can easily end up generating yet more legacy and hard to maintain legacy code in turn.

To take advantage of the benefits of a third-party approach (where someone else’s R&D dollars can be used to enhance functionality), most users would by preference go to market for an integration solution if it was an option.

It turns out that there are only a few software environments being positioned as useful primarily for the integration issue. Among these are IBM’s WebSphere, Microsoft BizTalk, Oracle’s Fusion and BEA WebLogic. But a player gaining more mind-share is 28-year old software specialist InterSystems with its Ensemble product.

"Though some users may prefer solutions more biased to an application server or Java Messaging Service approach, it is a good tool in this area, albeit from a small vendor"

Rob Hailstone, research director, IDC

Some readers may be more familiar with the Cambridge, Massachusetts-based firm’s Caché object-oriented or ‘post relational’ database, which has enjoyed great success in the global healthcare, telco and financial services markets.

A newer product which links with the former’s data structures is Ensemble, a universal integration platform that is positioned as enabling the rapid creation and fast integration of high-performance applications.
Working with Ensemble, claims the company, users interested in integration define business processes in an open and transparent way, then implement workflow and information integration and compliance, freeing business logic siloed in legacy systems.
Ensemble is claimed to be a platform that combines the functionality of an integration and application server, a high performance object database, plus a unified development and management environment in a single engine.

Gaining favour

Some commentators see the tool as a significant SOA solution in itself. Rob Hailstone, software infrastructure research director, IDC, says: “Based on a repository itself built in Caché, Ensemble works with the same metadata as the other product so it is well-integrated and uses a lot of the strong points of the other product, such as object persistence. Though some users may prefer solutions more biased to an application server or Java Messaging Service approach it is a good tool in this area, albeit from a small vendor.”

Ensemble is being used in a number of large-scale integration projects both here and internationally. At Florida’s Department of Children and Families, it is being used to integrate information from multiple disparate systems in a composite portal application that provides a single view of all relevant data about an individual. Some 150 applications will ultimately be connected using the InterSystems’ technology.
Also in the US, Partners Healthcare Systems is creating a front end using Ensemble to handle input data received from a number of member institutions, and then augmenting this information with enhanced processing and transformation logic to create uniformity of treatment of all its patients.

In Europe, Kimberly-Clark Europe in Brighton used a third party to implement an Ensemble solution that integrates software used by its in-house transport management group, over 100 small and medium-sized carriers with very limited technical capability and a central transportation management system.

To achieve this the firm has had to tie together a variety of communication mechanisms used by in-house and partner and supplier organisations, from fax, voice, flat file, spreadsheet and EDI.

This kind of integration project is an example of how users can not just integrate, but work to achieve not just ROI but return on existing investment.