Refer to the exhibit. You are designing this network to support voice communication between software and hardware endpoints. To ensure quality, the decision is to implement a DiffServ QoS model with…

Policing voice on switches enforces a rate limit and drops or re-marks excess traffic but provides no priority queuing guarantee, so voice packets still compete with data during congestion.

LLQ is a router IOS queuing feature specific to WAN interfaces; Layer 2 switches implement QoS through CoS-based or DSCP-based scheduling mechanisms and do not support LLQ.

LLQ (Low Latency Queuing) combines a strict priority queue with CBWFQ on router WAN interfaces, ensuring voice traffic marked EF DSCP is always dequeued first and never delayed behind data traffic, meeting the sub-150ms one-way delay and sub-30ms jitter requirements of VoIP.

Active queue management such as WRED manages congestion by probabilistically dropping data traffic early but does not create a strict priority queue and is not appropriate for delay-sensitive voice traffic.

Monitoring packet drops is a reactive diagnostic activity, not an admission control mechanism, and does not prevent voice quality degradation from occurring in the first place.