"In the end, Zdenek had a single-server CouchDB installation on the order of hundreds of megabytes of metadata stored."<p>I think the title is a little inflammatory, no?
Misleading title. It is only about a single person that used to work at CERN.<p>CERN has tons of developer groups, each with his own set of technologies.
Cost has nothing to do with it: they get it "free", as in US contribution in kind. And obviously, in the end of the day, Oracle profits much more.<p><a href="http://openlab.web.cern.ch/about/partners/oracle" rel="nofollow">http://openlab.web.cern.ch/about/partners/oracle</a>
Important note that I'm not seeing in other comments here: they started with <i>both</i> Oracle and CouchDB. This was basically a cleanup in favor of CouchDB, since running two different kinds of DBs is best avoided if possible for simplicity reasons.
Surprised it is not everyone else's reason: Cost.<p>Regarding database dumps, I'm curious how/why data pump[1] is not what he needed.<p>1: Overview of Oracle Data Pump: <a href="http://docs.oracle.com/cd/E11882_01/server.112/e22490/dp_overview.htm" rel="nofollow">http://docs.oracle.com/cd/E11882_01/server.112/e22490/dp_ove...</a>