Just as the world of services started drifting away from complex technology stacks and getting closer to simple, HTTP-based, possibly RESTful solutions, SAP have generated a new large, ambitious, all-encompassing specification named Unified Service Description Language (USDL). What is most remarkable of this approach is that it is not yet another WS-* technology or another process language. It is instead mostly focussed on capturing the main aspects that are relevant for carrying out business activities. This includes for instance pricing models, legal aspects, the notion of service bundles, the relationship between services and resources, etc. A “master data model for services” as they like to call it.
USDL is perhaps the most comprehensive specification around services which I have seen so far, bringing together economic, legal and technical aspects. Is it strange that precisely now that simpler service technologies seem to be gaining traction, they produce such a large specification? Perhaps. But one must also recognise that the technical aspects of services is not all there needs to be to allow the creation of service-based business solutions. In fact, for many services IT systems hardly come into play. These aspects have been dealt with by others previously, see for instance the e3 family of ontologies with which I’m a bit familiar. However, I am not aware of previous activities bringing all of this together in such a comprehensive manner with, presumably, the goal to eventually enable the creation of a platform allowing anybody to automatically trade (business) services much like the App Stores allows one to sell software.
The magnitude of USDL also carries as a consequence the fact that it is a large specification which will require significant work for its adoption. This is one of the reasons that has motivated the creation of the USDL Incubator Group within W3C. A remarkable choice of standardisation body which has actually triggered us joining in. Our interest, apart from our traditional interest on services, is centred precisely on trying to figure out how this technology could be brought for wider adoption on a Web scale. We do not even pretend that we will reach a solution that will ensure the adoption of USDL, that would be too pretentious. But I believe that working on trying to push out a specification of this sort to the Web apart from being worth the effort, it will definitely give us interesting insights on the dynamics of the Web when it comes to adopting specifications.
Our take on this, uses an example the evolution of the Web of Data, which has eventually managed to produce a larger number of vocabularies, most of them not standardised, and considerable amounts of data. I have therefore started working on producing a USDL version that is precisely driven by the main practices adopted for creating RDF vocabularies for the Web of Data, i.e., reuse, interlinking, simplicity, and why not, a bit of semantics. Eventually, we shall also illustrate how one could apply the Linked Data principles to publish USDL data. As I proceed I will try to report some of my findings and the decisions adopted in a family of posts. Let’s see if I can do some periodic blogging…