I recently was speaking with a senior IT manager at a major retail company, which is in the process of growing its market into wholesale–selling its goods to large companies, enterprises, and governments as well as individuals. The company had acquired a half-dozen firms that already made inroads into wholesale markets. It now was faced with the challenge of integrating disparate and dispersed information systems with the ultimate goal of creating a unified e-commerce system. Some of the acquisitions were on J2EE platforms with Oracle back ends; some were on .NET and using Microsoft SQL server; some were using open source and homegrown systems; and one still was stuck in the mainframe/midframe world with no current e-commerce.

My acquaintance was telling me how difficult it was going to be to decide on a common platform for all the entities involved. She suspected it was going to be a long and arduous process. Interestingly, she shared with me that the J2EE and .NET teams were open to change if it made business sense, but the open-source shop was vehemently opposed to adapting either of the commercial solutions.

I started to sympathize with her and then began to wonder whether maybe they were going about this in the wrong way. I would suggest it might make more sense to architect an e-commerce system that could interact with all, or most all, of the existing systems. Build an e-commerce system using any of the technologies available, and let that system pull and push data from the existing systems using Web services. Instead of forcing business units that have working information systems into expensive and unpopular change, why not retrofit the existing systems with a Web services layer. Hey, maybe some of these guys already had bought into service-oriented architecture (SOA) and are ready to rock and roll right now.