Totally agree — and the key nuance is: failover ≠ normal service. Rerouting usually prevents a full blackout, but the real pain is latency spikes, jitter/packet loss, and congestion, which can break “time-sensitive” workflows even if the app is technically online.
In finance/crypto that shows up as: delayed quotes, failed order submissions, logins/2FA timeouts, payment retries, and flaky API connections. Resilience isn’t just “more cables” — it’s diverse paths + capacity headroom + tested incident playbooks.