This is reminding me of last year's major Facebook outage. If I recall correctly, that outage was a bug in service discovery that took down all data centers (a

This is reminding me of last year's major Facebook outage. If I recall correctly, that outage was a bug in service discovery that took down all data centers (a CLI accepted a negative value when the Zookeeper variant treated it as an unsigned int , then all service discovery went down). I feel like service discovery is the biggest point of failure at large companies, and it would explain why services across so many different domains and systems went down.