A yo-yo attack is a Layer 7 attack vector aimed at elastic autoscaling, one of the failure modes a thorough DDoS test is built to exercise. The attacker sends a burst of load heavy enough to trip the autoscaler, then stops once new instances come online, waits for the platform to scale back down, and repeats. The victim's infrastructure oscillates up and down like a yo-yo, chasing a load that vanishes before the added capacity is ever put to work. The damage is economic as much as availability: the target pays for scale-out cycles it never benefits from, and latency spikes on every transition while instances warm.
Why it matters in DDoS testing
A yo-yo attack turns an availability control, autoscaling, into a cost and latency liability, so it is invisible to any test that only checks whether the site stayed up. A test characterizes the autoscaler's reaction window: how quickly it scales out, the cool-down before it scales in, and whether the burst period can be tuned to sit inside that hysteresis. It also measures the request-latency penalty during each scale transition and the spend accrued across cycles. Validating this belongs in resilience testing, where autoscaler behavior under adversarial load is the whole point.