Load Shedding
The deliberate rejection, degradation, or dropping of lower-priority work when a system lacks enough capacity to serve the full incoming load safely.
Resilience Context
Load shedding deliberately rejects, drops, or degrades lower-priority work when available capacity cannot safely serve the full incoming demand. Queue age, concurrency, CPU pressure, latency, and request deadlines can all be used as overload signals.
Policy Boundary
Random dropping is not always appropriate. Priority, idempotency, user impact, and retry behavior should shape the policy, and shedding does not replace long-term capacity planning.