...Major vendors first sold message queuing as the ultimate interoperability solution, then SOAP and REST, before realizing that multiple applications need to share data but had significant interface differences. Architects and vendors suggested many approaches to solving this issue, from writing wrappers using a common protocol to porting legacy systems to Java or .Net and, in the process, create a modern interface that fit in the enterprise architecture. None of these approaches is practical because they are code-intensive, expensive, and are coupled to specific systems, programming languages, and protocols......The solution to this problem is elegant and obvious: let the applications communicate with one another in the protocols they already support, from EDI to SOAP, over a common transport aggregator independent of the native protocols, and adding application- or protocol-specific translation modules or message routing only where required at the endpoints. This approach allows a mainframe application written in COBOL to interoperate with a mobile device written with J2ME without either end knowing anything about the other’s characteristics.
Ross Mason, a leading Java engineer, identified these issues as early as 2001 and began working on what became the Codehaus open-source project Mule. Parallel development occurred at Progress Software who produced Sonic ESB and coined the term “Enterprise Service Bus” during the same period. Several open-source and commercial ESBs exist now...
The case study then continues by looking at the deployment of Mule in an enterprise environment. After detailing a few tests of the setup Mule is looked at as a common JMS transport. The article then covers commercial support offerings for Mule from MuleSource.com. InfoQ has covered Mule and ESB's in the past including an online presentation on the topic from IBM's Mark Richards.