This letter coincides with the 2004 Software Development Conference in Santa Clara, where Ill be taking part in a Monday evening panel discussion on "The Marriage of SQL, XML, Web Services and Grids."
Moderated by the estimable Ken North, the session will offer viewpoints from Sun, BEA, Microsoft, IBM and Oracle in addition to whatever I can inject: Ill look forward, for example, to finding out what people have to say about the issues raised during the last such panel that I did with Ken, where transaction integrity and security model complexity were two of the high-profile topics.
The four-way "marriage" of the SD04 panels title is an interesting metaphor: Even a marriage of two technologies, such as XML and SQL or Web services and grid computing, is often enough to give birth to a houseful of choices and challenges.
For example, when XML and SQL get together, they may quarrel over treatment of recursive data relationships: At the end of this month, this conflict will be considered at the International Conference on Data Engineering in Boston. One paper scheduled for that conference promises "a generic algorithm to translate path expression queries into SQL in the presence of recursion in the schema and queries," which would be a significant contribution.
IBM, Hewlett-Packard, Computer Associates and others are among the wedding party for grids and Web services, which may exchange their vows after signing a prenup agreement defined by the WS-Resource Framework. That framework lets "stateful resources," such as the nodes of a grid, express their state in Web services formats.
When we get these happy couples together, in the ménage a quatre contemplated by our SD2004 panel, I shudder to think of the crockery-throwing disputes that may arise: For example, those fun-loving folks at Spire Security in Malvern, Pa., have already suggested that maliciously recursive XML payloads could become a vector of Web services attack. When computing grids provide adaptive capacity growth to handle such potentially explosive demands, its easy to see the potential for a concise but devastating assault.
Getting these issues addressed up front will ease enterprise concerns about utility computing, which IDCs Frank Gens has dubbed "futility computing" based on his perception that it wont soon be ready for prime time. Im not as pessimistic as Gens: There are plenty of real products saving real money for real users by applying utility computing ideas.
My optimism, though, depends on enterprise buyers asking the tough questions with expectations of answers, and not letting their questions go unspoken.