Lightsail workloads have a habit of becoming more important than anyone planned. A service starts as an internal tool or a prototype, quietly accumulates users, and ends up handling something the business depends on. The infrastructure review that would have caught a single point of failure never happened, because at the time nothing was at stake.

Nobody revisits these environments on purpose. There's no deployment ceremony to prompt a resilience discussion and often no runbook, so the first real assessment of the instance happens after it's already down and someone is trying to remember how it was configured.

Testing a Lightsail instance takes very little effort and answers the questions that matter for a workload in this position: how quickly you'd notice an outage, how the application behaves under load it wasn't sized for, and how long recovery actually takes. If any of those answers are uncomfortable, that's useful information about whether it still belongs on Lightsail at all.