When a new table is created, we're also defining the statistics on different columns (with Collect statistics on ... column ...).
These statistics are refreshed with the command Collect statistics on ... (without column option). If you look in the session monitor during execution, you can see that the table is scanned completely 3 times when 3 statistics were defined on the table (4 times with 4 collects and so on).
This is very time, cpu and io consuming. Is this 'problem' solved in Teradata 12? It's just a question.
It's not "solved" because it's not considered a "problem"... (works as designed ;) )
As far as I know, there was a feasibility study done on doing it in parallel but was later abandoned due to high resource consumption that would result from it. (if you monitor the resource consumption of a single stats collection on a huge table, you would get some picture)...