I also state that the INSERT...SELECT minimal logging isn't worth a hoot. According to a source at MS, it does "double writing" in the MDF file (that would be just as bad or worse than writing to a log file) and it still blows the log file space out if it's not already allocated. I just completed some testing for another purpose where the MDFF and LDF files were both started at 100MB with a growth of 100MB and the heap table met all of the conditions. The DB is in the SIMPLE Recovery Model. Doing a given SELECT INTO of a nearly 300MB table cause the MDF to grow to 300MB (as expected) and the log file remained at 100MB (as expected).After resetting the database to it's starting point, the same datasource was used for INSERT... SELECT. First, that took 45 seconds instead of the 7.5 seconds the SELECT INTO did. Many MS articles have stated that this would happen because it was never designed performance improvement (Seriously???). It also cause the LDF to grow from 100MB to 1600MB, which MS also documented as a "safety", which is really a left over of reallocation (paraphrasing their words for simplicity).And yes... this occurred with the proper use of WITH (TABLOCK) and, even though it's not supposed to be needed for a heap, I tried it with TF-610. Same results.I've not tested INSERT...SELECT for minimal logging so I don't know if it causes the same non-minimal problems (or results if you rather not call documented functionality a problem) but, if it hasn't changed, it's a worthless feature to me. No performance improvement and unnecessary bloating of the log file don't sound like anything really useful. Yeah... it might help backups a bit in the Bulk Logged mode but that's overwhelmed by the problems this feature exudes.Oh... and upon further reading, some say that you can't do big stuff with it or will result in the problems I just spoke about. I'll try that latter, as well but REALLY? It can't handle a piddley 300MB transfer?Doesn't seem at all useful and, if it has similar problems in 2012 (just got there recently), I know I won't be using it there, either. SELECT INTO and some of the other methods are much more effective for the amount of data that I do end up loading. I'll gladly take the final log file hit of the transfer of data from staging tables to permanent tables compared to the hits INSERT...SELECT has even when I'm in a SIMPLE Recovery Model for staging databases.

UPDATE:  SEE POST BELOW.