The first question is whether Oracle can give decent performance without huge efforts in tuning, whether in pure database administration or (possibly tuning-aware) SQL hand-coding. Unless the answer is “yes”, other ease-of-use considerations are too secondary to be dispositive.

In the past, there have been very few data warehouses above 5-10 TB where the answer was “yes”. Exadata purports to zoom the limit upwards. If that works out, the other advantages of Oracle at least have a chance to come into play.

See my recent post(s) on Oracle’s Exadata strategy for related thoughts.

CAM

“Note, however, that Gartner does not seem to assert that Exadata’s ease of use rivals that of the newer analytic DBMS specialists.”

==> that’s an interesting point. From a management’s perspective, having a large Oracle DBA team, I’d rather have my Oracle DBAs manage Exadata systems because they are already very used to Oracle tuning including parallelism rather than going for new technologies, even if they are “easy to use”. I guess many (most) corporate shops running DWH shops have DB2, Oracle or MS SqlServer expertise in-house.

This brings me to my second point: I am always wary of the “ease of use” concept. Ease of use, be it of a DB or an applicance always has to be weighed against control and features. The lack of both actually makes architecture design and production harder to manage and integrate in existing production frameworks (think of inter site replications for example).

Christian