Frank Cohen, now with Raining Data, has published an article about “FastSOA” on IBM developerWorks, the essential idea being that instead of turning XML into Java objects that are persisted to a DB, you should use an XQuery datastore in the middle tier.
I took a look at the Raining Data stuff a few years back and found it pretty impressive, even though they didn’t yet have XQuery then. It’s probably worth a look (and the mid-tier cache gibberish is obviously only intended to calm down enterprise architects who would NEVER EVER trust their data to something other than an RDBMS ;-)
(Oh, and what’s with the names? “Raining Data”? “Tiger Logic”? Come on … you can’t be serious.)
[via Loek Bakker]