Data Access: Service Data Objects
While not part of the SCA Model proper, Service Data Objects (SDO) are the de facto SCA data architecture. SDOs are based on the concept of disconnected data graphs. The SDO architecture maintains a tree or graph of data objects that can be accessed via APIs. The architecture allows both strong-typed and weak-typed data models, thus supporting static and dynamic data access strategies. The SDO architecture makes no assumptions about the query languages involved. Instead it relies on its base API to access nodes. Intended to separate application code and data access code, the architecture encourages the use of Data Access Services (DAS) to manipulate graphs (incubator.apache.org/tuscany/das_index.html). By adding metadata, SDOs also support runtime exploration of the data graphs. Since the architecture was designed with XML in mind, it provides a rich mechanism for manipulating XML. In fact, for practical purposes, the underlying data structure for SDOs may well be XML documents. At this writing, the SDO specification is in Version 2.1.
Conclusion
SCA attempts to create a simple, robust programming model based on SOA concepts. Its value lies in a conceptually simple, declarative approach to assembly and QoS. Through assembly-time and runtime wiring and loosely coupled policies, SCA addresses the issue of creating a service without foreknowledge of the underlying components. This, in turn, lets service component and integration developers act independently. Combined with SDO, a vendor-neutral approach to SOA is possible—the model designers appear to have learned from weaknesses of earlier models—by providing a clean separation between assembly and QoS. If the Apache Tuscany Project is an indication of SCA's viability, then the outlook for a straightforward (though not necessarily simple) way to implement SOA is promising (cwiki.apache.org/confluence/display/TUSCANY/Home).
