I am trying to think what would be the most useful demo of Vertica and decided to ask ... so ... what do all of you think?

How much data are you loading into each table? If you run DBD on the encoded table what encoding does it recommend for each column? Does the segmentation you have give an even number of rows across all nodes?

Not sure what is going on but do you have auto-commit on or off? What is your history retention set to?

But only if the other projections are truly "k-safe" for each other.

Try dropping them both in one drop statement.

I second power pivot over excel for even sets greater than 1m. When you run it, where is the bottleneck. Is it in the transfer of the data or what?

Specifically what tasks are you looking to do from the command line? Here are a few examples...

Start the database:
admintools -t start_db -d <dbname>

Stop the database:
admintools -t stop_db -d <dbname>