Companies’ information systems all share the same characteristic: change; they are permanently being adapted to changes in the companies they serve - and increasingly so.
The acceleration of this trend requires considerable effort from IT teams to meet needs for communication with partners, integrate functions offered by service suppliers, keep on top of the demands created by regulatory changes (Sarbanes-Oxley, Basel II, HIPAA, etc.) and handle changes in the competition, activity sector, and market. In addition, they must manage technological developments such as implementation of heterogeneous technologies. This balancing act must be performed while keeping control over costs, which includes improved team productivity, process streamlining, and software reuse.

The essence of SOA is to define independent (possibly contractual) business “bricks”, able to communicate with each other according to a pre-determined protocol. However, for maximum efficiency, they key factors are interoperability and the reusability of services.

So the fundamental points in an SOA approach are:
- Interoperability applies particularly to “visible” business services such as order entry, which anyone who has made an Internet purchase has already used. Easy communication outside the company is a must.
- Reuse is especially relevant for small, heavily-used unitary services such as calculations, a database management rule, authorization checks, LDAP access, etc. These elements remain relatively internal for the company, but are used by a number of different business services. Some may constitute a service layer for an infrastructure such as user management.
The various bricks constituting the information system will have specific needs. The goals of an SOA approach differ depending on the extent and scope of these bricks.
At the level of each independent application in the Information System, the main goal is reuse. Features that are repeated across applications should be ‘brought down’ to an Information System infrastructure level as a unitary service. Application-mining tools such as ARCAD-Observer will help you extract the business rules.

ILE-- particularly service programs that allow code reuse while improving application performance--and the ARCAD Open Repository, with its procedure repository--provides an entire range of tools, such as search by feature, independent execution of service program functions, and more, which optimizes the reuse of the functions created.

From the perspective of the overall Information System, we must isolate application groups, then determine their interfaces and transform them into services.
With its macroscopic views that allow both a top-down or bottom-up approach, the Application Mining provided by the ARCAD-Observer suite lets you navigate to a selected level in your applications to gain a clear and synthesized view of the Information System. The views highlight needs for application independence and interoperability.

Even if they don’t yet have a service-oriented architecture, existing Information Systems contain the business foundations that will serve as a strong benefit for companies. Moving them to SOA will multiply these values and accelerate company growth.

SOA is simply a type of architecture. You can think of it as a catalog of architectural concepts, or as a programming model.

In an SOA architecture, all functionality is provided as services assembled together by means of an application bus. This gives an unprecedented level of flexibility in deploying and reusing services.
The notion of the “application bus”, in other words the ability to make applications communicate with each other in a standardized fashion, is particularly appealing. Standardization is the wave of the future.

Just imagine: a new partner comes into your organization. Almost instantly, you can connect your order entry system to his, his billing system to yours--the increased efficiency for your organization is obvious.
Imagine integrating a new CRM software. It “plugs” itself in, using Web Services and existing management elements.
Result: No more double entries, no more developing interface programs, and instant overall coherence for the Information System.

ETL/EAI tools have already blazed a trail in this field. SOA goes much further, since users are no longer transferring and/or transforming data from one system to another. Instead, they are connecting processes. The challenge is to make your information technology communicate. After the standardization of network communications, where TCP/IP dominates the planet, it’s now up to applications to standardize the means of access to the services they provide.

APRIL Technologies’ need to modernize its Information System became clear about two years ago, with an AS/400 management application that was “getting old”. Projects around the application were becoming increasingly difficult to control, leading to escalating development costs. This modernization came hand in hand with a move towards new technologies for APRIL, eliciting essential questions such as: Which new technologies should be re-developed? Everything, or only presentation layers? What technologies should be chosen? What about business rules?

The need to start over again with the application had become unavoidable. So in 2005, the “Information Technology Plan” project was set up, including a “Functional Interoperability” track. As of then, an identification of the APRIL business function list became one of the key factors in the durability and profitability of the Information System.

However, technical choices had to be made before looking into functional rules.
After consulting concrete examples of the isolation of AS/400 business functions, very soon “ILE stood out as the best solution,” comments Gilles Gozlan.
Further, this functional integration involved centralizing functions on to a single machine, with the possibility of reusing them. The SOA approach was launched.
For the last two years, APRIL Technologies has been deploying business functions on its applications. These functions are reusable on the iSeries through ILE programming.
Once this barrier had been overcome, the natural course was to pursue this project by opening up access to the outside world (Windows and Unix platforms, etc.) to allow re-use of business functions by Web applications.

Since Web Services on System i are a recognized standard, they quickly won over APRIL Technologies, as they proved easy to produce and efficient to work with. With just a few clicks from the WDSCi wizards (WebSphere Development Studio Client for iSeries), you can create a Web Service to enclose the call for an ILE procedure without needing to know or program Java.
Once deployed in the WebSphere Application Server (WAS) on the System i, these Web Services are instantly “consumable” through a URL on any Web interface (.NET, PHP, Flash, etc.).
 
According to Gilles Gozlan, when it comes to renovating an Information System, APRIL’s approach varies depending on the circumstances. The projects are highly promising and have already reached cruising speed. As of now, the initial Enterprise Architecture phase is under control. To validate the Web Services section, APRIL Technologies decided to prototype the future “Insurance Clients” site for APRIL Assurance (these sites had been developed and hosted externally, but all future sites will be developed internally using .NET.).

Imagine a site written in PHP, which calls Web Services, which themselves invoke ILE System i business functions—and you’ll have a better understanding of the advantages of Web Services as a universal connector.
This is what APRIL is doing with its “Insurance Customer site” pilot project.

It’s amazing... but it’s clearly possible, and the proof is before you.

Comments were collected with the cooperation of Gilles Gozlan, technical architect and APRIL Technologies System i specialist.
(Note that in the SOA context, APRIL uses Software Configuration Management tools from the ARCAD-Skipper suite, which was expanded to the entire team in 2006).