Exactly. The org (and all constituent team members) needs to think about problems end-to-end. When you notice a problem, setup monitoring to collect metrics ab

Exactly. The org (and all constituent team members) needs to think about problems end-to-end. When you notice a problem, setup monitoring to collect metrics about the problem and alert on it before it becomes a downtime. Have ops write down their experience with that problem into a playbook, then have dev figure out which parts of the playbook can be automated, or if the system design can be iterated to remove the problem class altogether. If you're a dev and you don't listen to ops, you're just as doomed as an ops guy who doesn't read the dev's manuals.