time.time or datetime.datetime.utcnow() both give you a number that is independent of timezone (and datetime.utcnow() is supposed to be available as request.utcnow, although that still isn't the case).
However, as I pointed out, there is still a chance that this can go backwards in time on a production system. I've seen it happen before several times; This happens due to a misconfigured bios, a misconfigured ntpd, or generally misconfigured system on linux, and I-have-no-clue what causes it on Windows, but it does happen. Hence, all I want is a database counter (or fllesystem counter) I can increase. As I pointed when opening the thread, this is easy to do with the database, if somewhat wasteful -- I was just looking for a better alternative.
