Infinite Diversity in Infinite Combinations

"Buzzword compliance doth not a Platform make" but a primer on Service-Oriented Architecture (SOA) is in order to setup a common frame of reference and understand the myriads of acronyms like SOAP, WSDL, UDDI, etc that one encounters in this space.  

The basic tenet of SOA is that it is an architecture consisting of  loosely coupled "services" that are available and discoverable on a network. The exact definition of a “service” varies in literature and online standards but there is general agreement that a service is an application or software functionality that encapsulates non-trivial business logic and is able to interoperate with a variety of other services. In other words, a service is composed of intrinsically related business functions and data and only that service can manage that data. The interoperability requirement dictates that service providers “publish” information about them in a central registry, where other services that need to consume their service can "discover" them and then communicate.

This “Publish-Find-Bind” triangle, with a central registry of services,  is a common pattern in SOA however there can be other methods of discovery. Most notably, as John Robinson describes, our Open Client Platform (which is essentially client-side SOA) consists of aruntime service called the Execution Environment (EE) which does away with the central registry idea and simply queries all the service manifests installed on the client machine. For most host-side services, however, a central discovery data repository remains the common pattern of choice. There are several vendors out there with service repository products -- I recently got an email from Systinet evangelizing their service registry product.                                            

More to follow on SOA...

|  Add to del.icio.us |  digg this