The internet was built to survive failures, and that philosophy runs through every layer of its infrastructure. Major data centers maintain at least two independent power feeds, backup generators, and redundant cooling systems.

Network traffic follows the same principle. If one undersea cable breaks, traffic can shift through alternative paths, but the timing is not always milliseconds. Local fast-failover systems can react quickly, while global BGP routing changes often converge over seconds or minutes.

This redundancy is expensive but essential. When a single AWS region went down for several hours in late 2021, it knocked offline services ranging from smart vacuums to airline check-in — a reminder that "the cloud" is just someone else's very physical, very breakable infrastructure.