Good:
at 10:30 AM, one of our back end message servers, serving roughly one third of the user population, died. no worries — it seamlessly failed over to its partner, which picked up the phone, and life went on.

Bad:
at 1555, support started to get calls because webmail on one of the other back end message servers, also serving roughly one third of the user population, inexplicably died. five minutes on hold (and my GOD mirapoint has some bad hold music, and i don’t mean bad as in “cool”) and the engineer was like, “well, that shouldn’t have happened!” well … no, actually, it shouldn’t have, thanks for saying so. twenty minutes down. whee.