It started innocently enough with a gs.info(). But when the syslog table refused to load, I knew something was wrong.

2 million, 4 million, 8 million logs per day, I've seen it all and while volume matters, the real question is:

How do we reduce them?