One of my continual gripes with the DBCT is it's poor behavior on large data sets. See NI's response here:
http://digital.ni.com/public.nsf/3efedde4322fef19862567740067f3cc/862567530005f0a186256a33006b917a?OpenDocument
Actually, I wouldn't find it nearly so annoying but for the part that says "In most database books you are told not to return large recordsets...". I prefer the freedom to make a mess of things myself, thanks.
On the plus side, the VI referenced in the article is fairly efficient, you just have to deal with type conversion once things have been moved into LabVIEW.