Say your startup has an on-call rotation that follows the sun. Someone’s always on. One day, a member of your team gets an alert that disk space will run out on the primary PostgreSQL server in two hours. An hour later, he gets another page. Then thirty minutes later, disk space runs out and your entire site goes down. For 18 hours. The silver lining is that your two-tier architecture can keep your site and app online, but the backend of your business has come to a halt. Your CTO is furious. He says you must fire the engineer who was on call to make a point.
