A Case for Enterprise Application Integration

Enterprise application integration (EAI) should not be viewed as a destination but rather as a journey. Higher business value can be realized as increasingly complex business processes are automated, standardized, reused and shared. Yet the costs of enterprise application integration can also be substantial, both from a financial standpoint and in terms of the organizational disruptions that are often involved. For that reason, building a strong business case is critical to obtaining the organizational support that is required for a successful implementation. This article will provide a detailed case history that demonstrates how EAI can save money, not only over the long run but even on the very first project on which it is implemented.

Virtually all global 2000 companies are planning major information technology initiatives in the next three to five years. Some of the most popular include web-based commerce, business-to-business commerce, customer self service, expanding or integrating enterprise resource planning systems, mergers or acquisitions, compliance with major regulatory changes, business intelligence initiatives, supply chain integration and mobile computing. Nearly all of these new systems will run on the data in today's legacy systems, which means that they will all require some form of integration. The danger lurks that these integration issues will be addressed one at a time as an integral part of each project, with the result that the costs of legacy integration will be incurred over and over again. Many applications, in fact, will bring middleware to address their specific integration needs. This means that, in the absence of an EAI strategy, multiple EAI solutions will arise almost spontaneously, with each requiring their own hardware, implementation costs and maintenance.

EAI presents a systematic approach to this problem by delivering solutions that are capable of ushering volumes of complex transactions through entire business processes across multiple platforms and systems. Using process modeling, companies can define potential solutions, simulate solutions to analyze flow of information and easily change the process with minimal coding. Vendors that focus on business or process automation include Tibco, SeeBeyond, webMethods, Vitria and others. This approach challenges organizations with integration on a scale that they have never attempted before and requires a correspondingly large infrastructure investment to ensure integrity of mission-critical transaction processes. But, despite the upfront investment in architecture, an EAI solution is usually less expensive than custom integration because it provides an infrastructure that simplifies the creation of and allows reuse of application interfaces. Concentrating the data transformation logic into a single environment also makes it much easier to learn and manage.

Costs of integration
EAI costs come in three components: architecture, integrations, and operations. Architecture costs are capitalizable costs related to the initial deployment such as integration development, execution and operations environments. They include the license cost that is negotiated with the vendor, the cost of new hardware required to develop, run and monitor integrations and the cost to implement architectural software and hardware. Roughly 80% of architecture costs are incurred within 6 months of implementation while additional expenses may be incurred for hardware or licenses as usage spreads. Architectural costs are driven by the complexity of the EAI software and the number of discrete business entities to which it is deployed. As you might expect, the architectural investment is much higher for an EAI solution than custom integration.

Integration development costs are separate from the architectural costs. Integration costs are often capitalized and relate to the development of interfaces and collaborations between systems. Integration costs are variable and are driven by the number of interfaces that are developed. Integration costs with EAI are generally between 25% and 40% lower than with custom integration. Development is less expensive because adapters come pre-built with the EAI architecture and the architecture provides a graphical interface in which to perform mapping as well as many pre-built functions. Examples include message transport, guaranteed delivery and process control including the ability to re-use process steps. Another important point is that since all applications communicate with a common middleware, far fewer interfaces need to be developed. It's also important to note that the EAI software contains functionality that makes developing, running and monitoring the integration system much easier. This includes business process management facilities, reusable transformation and formatting components, auditing, logging and debugging functions; pre-built adapters with upgrades for future releases of software, message receipt acknowledgement and much more.

Operating costs are expensed and include on-going operations and maintenance of the EAI system for architecture and integrations. Operating costs generally are driven by the number of interfaces that need to be maintained and rise as more interfaces are put into production. EAI generally provides a 50% to 80% reduction in application maintenance cost by reducing the number of interfaces that need to be maintained and offloading much of the costs of interface maintenance onto the EAI solution provider.

Typical comparison of custom integration vs. EAI
The economic benefits of EAI are so significant that it can often be justified in a single project and provide substantial additional benefits from that point forward. Consider a global 2000 company that was planning to integrate its ERP system with a business portal, a project that required a total of 190 interfaces. As shown on the accompanying chart, architectural costs for an EAI solution based on CrossWorld's software amounted to $1,630,000, including $700,000 for software licenses, $90,000 for hardware and $840,000 for architectural implementation. The cost of architecture for the custom solution is only $802,000, since software license fees are eliminated and implementation costs are lower. Custom integration architecture costs are made up of the following activities: periodic processing script design, build, and testing; job scheduling script design, build and testing; built in transaction control design, build and testing; reusable interface frameworks such as request/reply for portal interfaces; data translation program design, build and testing; and data extract program design, build, and testing. It is not uncommon to find these costs hidden in the development of the individual interfaces.

The primary EAI advantage comes in the area of interface development. Each custom interface takes 6.75 days to analyze and design, 18 days to perform detailed design, build and test and 7.25 days for system test for a total of 32 days. At a blended application development rate of $1000 per day, this amounts to a cost of $32,000 per interface. Designing interfaces for reuse takes 30% longer so analysis and design time is 8.78 days for the EAI approach. On the other hand, the EAI product used in this application, CrossWorlds, provides a 25% productivity gain in development relative to the custom approach, so detailed design, build and test time is reduced to 13.50 days. System testing time remains the same at 7.25 days. Total time to build each interface adds up to 29.525 days for a total cost of $29,525. An even more important difference with EAI is that 42 of the 191 interfaces involved in the project can be reused. Of course, re-use isn't free, but on this project interfaces that were re-used were 80% less expensive than the ones that had to be developed, with the remaining cost primarily consisting of systems testing time. The net result, summarized in the accompanying chart, is that the architecture and integration costs for the EAI solution are $488,955 less expensive on this single project. When maintenance costs are considered, the EAI advantage will be even greater.

Of course, on future projects, with the architectural costs of EAI already in place and the potential for even greater re-use of interfaces, the cost advantages of EAI are substantially higher. This company's second project, which involved upgrading one ERP system and integrating it with another ERP instance on a different release, involved a total of 100 interfaces of which 22 could be re-used. The difference in architecture cost in this case was considerably smaller because the EAI architecture was already in place. The custom architecture development cost was $420,000 while the EAI architecture cost was $445,916 due primarily to licenses and hardware necessary to integrate the new ERP instance. The cost per interface, on the other hand, was the same as in the first project, $32,000 for custom and $29,530 for EAI adapters. But, 22 of the EAI interfaces were reduced to a cost of $5906 because it was possible to re-use existing interfaces. As one would expect, a conservative approach was used to calculate reused interfaces. The potential interfaces reused within the second project and from the first project could be much higher. The result was that the interface development costs on this project for the EAI approach were only $4,895,245 compared to $6,112,000 for the custom approach, and total integration costs were only $6,425,245 for the EAI approach compared to $7,014,200 for a custom solution. The EAI cost advantage continues to grow as additional projects are implemented, as the accompanying chart shows.

It is important to note that while the business case indicates that EAI is justified in the first project, we continued to look for other projects that might use it in order to increase the certainty of the decision. The learning curve associated with EAI will make the initial integrations cost much more than future integrations. Therefore, it may take more than the projected 110 interfaces for the development effort to break even.

EAI also provides other advantages that are difficult to quantify. The increase in development productivity shortens cycle times for development and frees resources to support further integration efforts. The flexibility of EAI allows for changes to the business and technical landscape to occur with minimal rework and impact on production systems. Increased manageability and maintainability provides extended technical control of the environment for proactive and reactive management of systems. The EAI approach also provides improved access to timelier and more accurate data across a distributed environment while minimizing redundancy. Finally, business process management promotes the management of integration at the business process level and allows for real-time and historical analysis of business conditions and performance. The bottom line is that EAI is a strategic investment that can reduce costs, not only over the long run, but often on the very first project.

*Please note this article uses examples only. Actual savings from using EAI and whether such savings occur can differ based on a wide variety of factors.