On Huawei OLTs, the standard QoS pipeline for upstream packets is: first perform traffic classification (identify flows), then marking (set CoS/DSCP as required), next apply traffic policing (rate-limit and color traffic), followed by congestion avoidance (AQM such as RED/WRED), and finally congestion management (queue scheduling/priority queuing). This ordered sequence ensures correct identification and marking before any rate enforcement, and that packets encounter AQM prior to scheduler-based dequeue.

Type C dual-homing provides device-level protection with two OLTs; when one OLT/uplink fails, services switch to the peer OLT. In single-homing, ONUs are associated with only one OLT; therefore an OLT failure interrupts services (C). Single-homing has simpler networking and management than dual-homing (D). It does not offer switchover to another OLT (so B is incorrect), and the “important-service” positioning in A is typically applied to dual-homing scenarios that require the highest availability.

In capacity planning, the planned bandwidth is computed as:

Planned Bandwidth = Users × Penetration × Concurrency × Online Bandwidth × Duty Cycle × Channels × Coverage.

Substituting the values: 500 × 0.5 × 0.4 × 4 Mbit/s × 0.125 × 1 × 0.7

= 250 × 0.4 × 4 × 0.125 × 0.7

= 100 × 4 × 0.125 × 0.7

= 400 × 0.125 × 0.7

= 50 × 0.7

= 35 Mbit/s.

This matches the given result, so the statement is true.

Huawei devices support independent mapping/remarking for inner and outer VLAN priorities (802.1p CoS) when QinQ is used. Operators can define distinct mapping tables or remark rules per tag. Option A is incorrect (priorities can be derived or remarked, not “must be” tag-in-packet), option C is incorrect because mapping profiles are configurable, and option D confuses VLAN IDs with priorities; inner/outer CoS can be mapped independently for inbound/outbound directions.

In Huawei’s QoS theory (aligned with IETF), the recognized service models are Best Effort, Integrated Services (IntServ), and Differentiated Services (DiffServ). FIFO is a queueing discipline (scheduling method) rather than a service model. FIFO simply forwards packets in arrival order without classification or guarantees, and while it can be used within Best Effort, it is not itself a service model.