XML Data modeling for Web publishing workflow

Tasked with consolidating numerous Web publishing systems into a single enterprise architecture, The Content Management Engineering (CME) department at Sun Microsystems is responsible for the core of the company’s global Web presence, http://www.sun.com . CME publishes a variety of information from a diversity of sources, so in order to streamline workflow for complex business processes, and to maximize reuse of content, the department has developed a library of source content XML, governed by an XML schema portfolio. This schema is driven by an underlying data model, bound to content elements through annotations in the schema.

The data modeling combines elements of classic enterprise data architecture with developments from the age of Web and agile development. It has to connect XML schemata to other systems ranging from ERP and traditional data warehouses to stand-alone relational databases to N-Tier enterprise Java systems to flat file collections. The problem space is complicated by business factors such as globalization, e-commerce features, and large changes to many of the systems that feed CME. This opens up many problems for data architects. Some of these (versioning, governance, etc.) are well discussed in the industry and some of them required CME to chart it’s own suitable practices, and often to look for delicate compromises. One example of such problems is how to deal with modeling of content fields that have no natural length limitations, but which may be constrained by implementation limitations in legacy systems.

This presentation outlines these modeling techniques that have proved successful as scope of the web venues and responsibilities of CME have steadily increased. It discusses the more interesting problems for data modelers and XML professionals raised in CME’s projects.