EAI: Integrate, mate

Web services

The connectors and adaptors currently used for EAI could be replaced by Web services, “but how you control all of that interoperation and integration is still something that Web services doesn’t satisfy,” he says, as you still need to define how the various applications fit together. “It’s a while off until we see applications consuming information delivered by Web services,” he suggests.

McCabe’s opinion is similar; although the idea of Web services is “seductive”, he expects they will be used mainly for internal integration during the next three years. While the idea of a platform and application-independent technology is appealing, “nothing is a done deal” as a key player could decide to take a different direction.

One of the key players is obviously Microsoft. According to a company spokesperson, “BizTalk Server unites EAI, B2B, and business process management technology in a single product to allow companies to easily orchestrate Web services and rapidly build dynamic business processes that span applications, platforms, and businesses.” Over 1100 of its customers including Ford Motor Co. and the UK government are already using BizTalk Server for EAI.

A wide variety of adaptors allow BizTalk Server to receive “documents” (bundles of information, not necessarily documents in the desktop computing sense of the word) and after transforming them, send them on to another application. These adaptors work with applications such as Hogan Financials, J D Edwards, Oracle Financials, PeopleSoft, SAP, and Siebel, or with technologies including CICS, COM, CORBA, dBase, EDI, FTP, Ingres, Java, Oracle, SQL Server, and a host of different green screen terminal types. Furthermore, BizTalk Orchestration takes care of the workflow aspects by automating the business processes that the EAI project is intended to support.

Plank says that unlike other EAI products, BizTalk Server “is not just for large organisations.” Other vendors offer “absolutely fabulous products, but [they are] priced for the enterprise,” he says. With BizTalk Server, “it’s not just the licensing costs that are low, the consulting costs are too.”

van Gore takes a pragmatic view of Web services: it is easier to save a dollar by improving internal systems than to try to make a dollar by selling a Web service. He believes the introduction of Web services will be slowed by the state of the economy, and that organisations will concentrate on internal EAI. Web services is a medium to long-term winner, he suggests.

This meshes with Toohey’s experience: he has seen “a lot of genuine interest from customers and prospects” in Web services, but more for EAI purposes than to deliver services outside the organisation. He noted there is less local interest in Web services than there is in the US. J D Edwards plans a staged introduction of Web services support. It has started by delivering standard functions as Web services, and will add the ability to publish custom functions. Next will come Web service consumption, and finally a supply chain optimisation engine delivered as a Web service.

Marriott suggests using new technologies such as Web services for EAI fails to change the fundamental business processes which the technology is designed to support, and leaves the organisation with a patchwork architecture. “Oracle’s technology integration solution provides a complete platform to allow enterprises to integrate the new with the old. Web services alone won’t solve this very large problem,” he says.

Software that puts a wrapper around legacy systems makes it easier to replace components as required, says Knochs, but when it comes to choosing Web services or another technology, “it’s up to the customer to make that decision on a business basis.”

Exposing information through Web services leads to the idea that one core application could act as a Web services consumer, drawing in all the information that users need from other applications. “The promise of Web services is that when you upgrade an application it rediscovers and reintegrates the services available [from other applications], but that’s a long way off,” says Menadue. This isn’t possible yet, and he warns that this arrangement could make it difficult to replace the core application, so it might be better to abstract information away from individual applications (eg, via a portal) so that any of them can be replaced when required.

“Web services is very elegant, but you need a business case for the elegance,” says Kloss.

Portals

“The portal paradigm for presenting information is a good one,” says Menadue, but he suggests you shouldn’t think exclusively of Web portals. “You don’t want to repeat that integration work” if you subsequently add a speech portal or a WAP (or other wireless technology) portal, so the portal should be treated as “another layer of presentation abstraction” on top of a services-based architecture. “The challenge is sorting out the right way of implementing a portal,” he says.

The basic idea of a portal is that it provides uniform access to information and applications. Dimension Data’s internal portal integrates sales, inventory, project, and customer information that is collected from enterprise applications such as SAP and Siebel as well as in-house programs. Ultimately, users don’t care where the information comes from, Menadue explained. This also provides an opportunity to deliver server-based applications (via the portal) rather than deploying software on each desktop system.

A portal can be a great way to link people and information, says Kloss, “but it doesn’t give end-to-end business processes” so you must identify the workflow involved.

How do you decide whether a portal is more appropriate than integrating other software with a master application? Toohey suggests that portals are good for situations such as providing senior executives with key performance indicators drawn from multiple systems such as CRM and ERP. But where a group of users spend most of their time in one particular application (eg, call centre agents and CRM), it is better to leave them there and implement behind the scenes EAI that exchanges information with the other systems such as ERP.

Legacy systems

A decade or two ago, we saw a move from host systems to client/server, but what we actually got was hosts and client server, suggests Kristin Munger, market manager at WRQ. That picture has become even more complex with the addition of ERP and similar systems, plus Web-based applications.

WRQ’s Verastream can take logic and information from a host system and publish it as a Web service or as a component so it can be reused in distributed applications. This has two benefits. Firstly, Verastream “is non-invasive on the host side” as Munger puts it, so any shortage of host development staff is not a problem and current skill sets with component software and Web services can be put to work on the project. Once the component repository has been populated with an inventory of the logic available on the host systems, these components can be reused as required.

Secondly, existing applications can be deployed in new areas instead of being replaced with new systems. She gives the example of a large insurance company that wanted to deliver a host-based risk management application to a new group of users that had experience with the traditional green screen interface. By using Verastream, the company was able to deliver information collected from several applications through a single interface. This approach also means that any of the underlying applications can be replaced down the track without affecting users.

“A technology that lets you prolong the life of an application is an advantage,” says Knochs, but McCabe counsels that putting a wrapper around an old application is “always harder than it seems.”

A major driver for EAI can be mergers and acquisitions. The cost of converting from one company’s chosen system to the other may be prohibitive, but in order to find the economies of scale that drove the merger you have to do something. Thew pointed to a merger between two Japanese oil companies that used Interstage to integrate the two distribution systems—the result was a US$300 million saving.

Kloss says plenty of examples can be found in the finance industry, where banks have merged and purchased insurance companies. One bank has 19 separate customer information systems.