Browse by year:
March - 2003 - issue > Cover Feature
Next Generation EAI
Evangelos Simoudis
Thursday, February 27, 2003
ENTERPRISE APPLICATION INTEGRATION HAS emerged as a distinct and important layer of the enterprise software infrastructure. Packaged EAI software can be deployed—in place of several manual processes and one-off connections––and allow an enterprise to, for example, integrate its marketing automation, ecommerce, order processing, and billing applications thus creating an end-to-end customer management system.

Today application integration ranks among the top three IT initiatives of most any Global 2000 enterprise. However, for most enterprises, EAI remains an expensive and time-consuming proposition resulting in fragile systems. Therefore, EAI tools and methods are still not used widely, even among the largest enterprises where most EAI projects consist of establishing basic point-to-point communication between departmental applications. CIOs recognize the need for a new generation of EAI that will effectively address the shortcomings of existing EAI solutions.

First-generation EAI
EAI emerged in the mid-1990s with the founding of companies such as Active Software (now part of WebMethods), Vitria, Crossworlds (now part of IBM), etc. These companies recognized that application integration represented a significant IT pain (even today enterprises spend over 40% of IT budgets on custom programming to tie together applications) and enterprises desperately needed a systematic way to tackle the problem.

Early adopters of EAI––companies like Motorola, Shell and Morgan Stanley––realized significant return on their investment in first generation EAI tools. This ROI was attributed to the fact that the companies no longer needed to employ a cadre of programmers just to create, maintain and support the homegrown integration software.

Like every other software technology, packaged, tools-based EAI, despite its unequivocal success, is not without its shortcomings:

It’s expensive––Initial EAI tools consisted of a few application connectors (or adapters) and simple message routers that performed basic application integration functions. However, over the years EAI tools have evolved into multi-layer stacks with rich functionality and an increasingly higher price. Today the average license cost of such integration stacks reaches $1 million.
Expensive to deploy––For successful deployment EAI tools require significant customization in two areas. First, there is the need to “link” appropriate data elements of any two applications that must be integrated, in order to enable the flow of data between those applications. For example, if the customer identification number in the billing application is called “CID” and in the CRM application is called “Cust-ID” it is important to link these elements if customer identification data is to flow between these two applications. Second, each adapter must be “tuned” in order to optimize the performance of the integrated system. Because of the intricacy and criticality of the linking and tuning operations, corporations employed system integrators to perform them. The nNet result is that system integration consultants replaced programmers employed in the past––or more bluntly––application integration using first generation EAI tools replaces one type of expensive human resource with another. System integration costs usually balloon to 3-5 times those of the first generation packaged EAI software license costs. Furthermore, these costs continue to grow almost linearly as new applications are incorporated into the initially integrated set.
Slow to deploy––Even small-scale EAI projects using state-of-practice tools can take months to deploy.
Based on proprietary architectures and technologies––Present generation EAI tools are built on closed, proprietary technologies. This makes it difficult (at times impossible) and expensive to take full advantage of emerging Internet-based data representation and communication standards, e.g., message-passing based on SOAP, and to replace parts of proprietary EAI stacks with components that are standards-based.
Provide limited ability to integrate the business with the IT side––Present tools are only barely capable of integrating business processes with the appropriate applications that must be invoked in order to execute these processes. In other words, application integration based on first-generation tools has barely risen above physical integration of software systems.

Next-generation EAI
Today, a full seven years after the first EAI company was founded, over 30% of Global 2000 enterprises do not use any packaged EAI solutions. Moreover, only a fraction of companies that use EAI solutions have deployed them enterprise-wide. And while EAI continues to represent a significant IT pain, none of the current market-leading EAI companies commands over 40% market share. Existing EAI market leaders trying to address the identified shortcomings cannot do it fast enough––sometimes because they cannot adopt new technologies quickly and other times because their business models cannot support the necessary changes. Finally, CIOs believe that Internet standards evolving around XML, e.g., web services, etc., will address the identified shortcomings.

These observations lead to two conclusions. First, the EAI market will continue to grow––fast. For example, this year the EAI market is sized at roughly $5B and is expected to grow to $11B within the next three years (IDC). Second, a new company can emerge as a leader because there is still a large unaddressed market and the incumbents cannot claim it fast enough. Therefore, investors are particularly excited about building the companies that bring to market the next-generation EAI solutions.

Characteristics of Next generation EAI solutions
• Inexpensive to acquire, maintain and support with easily demonstrable ROI and low Total Cost of Ownership (TCO)––Global 2000 companies are replacing existing EAI solutions used to integrate customer-facing applications with new, standards-based solutions in order to halve the TCO of EAI.
• Fast to deploy––Next-generation EAI solutions typically take days rather than months, requiring no more than a 2:1 ratio of system integration services to license cost.
• Standards-based––New applications take advantage of Services Oriented Architectures, the evolution of existing Internet-based standards such as XML and JCA, and the incorporation of new standards as they emerge, e.g., XQuery, WS-Security.
• Expanded notion of integration––Today’s EAI solutions are limited to integrating enterprise applications. As the enterprise becomes distributed but highly interconnected, next generation EAI solutions will allow for the integration of data sources such as databases, files, data streams coming from devices that reside at the edge of the network, etc., in addition to applications.
• Enable the speedy integration of data sources and applications ––Ultimately, new applications and data sources that are “wrapped” as web services, i.e., have standards-based APIs, can be automatically detected and integrated into a solution using Services Oriented Architectures. This area, now called “composite application development,” is quickly emerging as several small companies have gained funding over the past six months.
• Make explicit data and application semantics––The meaning of each data element and of the relations among the data elements must be represented explicitly. The manageability of EAI solutions will increase and the costs will decrease.
• Capable of fast information processing speeds to enable the real-time enterprise––As the integration among enterprise applications and data sources increases, corporations are interested in facilitating the real-time flow of data to take advantage of emerging opportunities or avert impending crises.
• Secure and able to provide business continuity––One area where existing EAI solutions indisputably excel today, as with most proprietary solutions, is in providing strong security and business continuity.
• Capable of integrating business processes with IT systems––In order to better control the enterprise it is particularly important to integrate business processes with the systems that run them. For example, SAP is developing xApps, a series of applications that use web services standards to integrate business processes with applications. Next-generation EAI systems incorporate monitoring and management capabilities that provide “enterprise visibility,” i.e., provide the user with the pulse of the enterprise and allow business objectives to be effectively aligned with IT resources.

Share on LinkedIn