A service's average latency looks normal while some customers experience severe delays. What should be examined?

1. The length of the resource's display name.
2. Only the compartment's budget threshold.
3. Only a longer monthly average.
4. Tail latency, error rates, workload dimensions, and traces for affected requests.