What metrics do you prioritize when defining service health for a critical infrastructure component at Slack?
