Lambda logs show occasional high 'Init Duration' spikes while most invocations stay fast. What's happening and what's the mitigation?
A.Cold starts — initialization runs on new environments; use provisioned concurrency or reduce init code
Init spike = cold start; warm pools or lighter init fix it. Correct.
B.Memory exhaustion — increase the timeout
Timeout doesn't address init latency.
C.A bug in the handler — add retries
Cold starts aren't bugs — retries don't help latency.
D.Throttling — request a concurrency limit increase
Throttling shows different symptoms (429s), not Init spikes.
Why this answer
Init Duration = the init phase running on a NEW execution environment (cold start) — code loading, SDK clients, connections. Provisioned concurrency keeps warm environments; lighter init (lazy imports, connection reuse) reduces it.
Correct answer: A · Last reviewed 2026-11-05
Official AWS references
Practice this interactively
Answer without seeing the solution first in the Debugging & Optimization practice session.
Practice Debugging & Optimization