RedMonk
putting technolgy strategy into context
AboutClients Expertise PressBlogLibraryHomeContact
  Web ServicesApplication DevelopmentApplication ManagementApplication Servers
Enterprise Content ManagementEnterprise Application IntegrationEnterprise Portals
 
 

Enterprise Application Integration (EAI)

It seems as if nearly every IT market is "in transition" right now and unfortunately for its proponents, EAI is no exception. What is EAI and why is it important? From ERP to CRM, from SAP to Siebel, from HR to Manufacturing, from Peoplesoft to JD Edwards, every new packaged application that comes down the pike promises massive advantages but initially fails to deliver. Why? Because the new stuff doesn't play nice with the old stuff. Integration is IT's biggest challenge but not enough organizations are systematic about their integration architecture to meet the challenge.

EAI was introduced as an answer to this problem, to overcome some of the limitations of piecemeal integration. The pitch from EAI vendors was that, rather than tying together systems by hand, users should buy connectors and services in the same way they bought other software--by licensing packages and paying for pre-built connectors. However, this was still done in a point-to-point fashion. Each application was tied directly to another. The problem with this approach was that it didn't scale. Vendors therefore introduced a more systematic approach and began to sell integration hubs--providing a hub and spoke architecture for integration, a central point for plugging in and connecting disparate applications and platforms. Thus the EAI industry was born. Proprietary tools and methods ruled. Microsoft plays in this world through its BizTalk server, which is of course being rebuilt as a Web Service-based platform as you read this.

Today, however, standards-based approaches are fast taking over from traditional EAI and the world of proprietary integration is looking a little stale. The rise of Java and XML Web Services is changing the dynamics of the integration tools market, prices are falling, and vendors failing to respond to market changes are being eliminated from the shortlist for request for proposals (RFPs). Anyone working in the field knows that legacy integration mechanisms are in many cases more scalable than standards-based mechanisms--they have had time to bed down and vendors have spent years investing in high performance messaging to ensure the tools scale in the most demanding of environments. In markets like financial services and telecom, scalability is not an optional extra. But standards-based approaches are catching up and they are far cheaper.

The other major change facing EAI is that is e-business has changed the way organizations want to connect to each other. Business to business integration must be more flexible than ever, as business partnerships are made and broken, and merger and acquisition activity constantly changes the business landscape. The problem was that EAI vendors focused on internal integration while new entrants, such as Extricity and ActiveSoftware (purchased by WebMethods) focused on B2Bi. Thus silos were created and users were buying from one set of vendors for internal, and another for external, integration. Hugely ironic that, given that EAI was supposed to overcome the silo problem in the first place. Something had to give and the market started consolidating rapidly as vendors tried to stitch together new companies out of seperate ones specializing in different models. In some cases consolidation was too rapid--Peregrine's bankruptcy can be squarely pegged to its attempts to become a full service provider of every kind of integration on top of its traditional asset management business.

XML Web Services is changing the game in numerous ways--for one thing it offers a single common standard approach to integration, whether internal or external. For another, because it is based on open standards, prices are by definition low--after all, standards means competition on the basis of implementation. Without lockin it is awfully hard to justify premium pricing.

RedMonk believes users must be more systematic in their integration strategies. Web Services is not an instant replacement for EAI. It's not that simple. But users should begin to familiarize themselves with newer integration models so they can manage the coexistence and transition between the two. But being systematic means far more than choosing an integration mechanism. The fact remains that organizational change is the key to successful integration. Tools can underpin success but first the organization needs a mindset and structure that allows for effective integration. EAI can break down technical silos but not operational ones.