Cart
[Home] [By Thread] [By Date] [Recent Entries]
Before we pull to hard to the RESTstop, one thing. This quote from Michael Kay's article prompted: "I hope I've encouraged you in this article to recognize in particular that XML databases are not just a direct replacement for relational databases. They fulfill a different role in the architecture, and they work well when supporting an application designed around the notion of data on the move rather than data in the warehouse." Strictly speaking, in a workflow app that is moving documents from node to node, this is true because documents are the predigested bits of beef where the formal act of negotiating the negotiating instruments is done. (Lou Burnard: "A DTD is a theory about a document") On the other hand, when doing analytics, warehousing is a good idea because the whole point is to theorize on the data. That is why business intelligence systems (see Cognos, XML for Analysis, etc.) emphasize ad hoc querying, hypercubes (mapping into a predigested theory of theories) are emphasized. The network wants as little semantic as it can handle to do it's job. The data analyst wants to create semantics or prove them (see pragmatics) and the marketing exec wants to appear as if they are cooperating but in serving the stockholder/stakeholder values, want to find ways to confuse the other guy's theories (see combative pragmatics). Thus, economies are NEVER frictionless. They thrive on it because it keeps the spice in motion instead of settling in a sub-optimum valley (noise in service of annealing). len
|
