QoS monitoring, assurance and troubleshooting
Scope
This domain tests how you prove QoS is working and isolate faults on Nokia routers and access nodes.
Core rules
- Use operational commands:
show qos sap-ingress,show qos network,show qos scheduler,show qos slope,show port,monitor port,show service sap statistics,show service sdp statistics. - Queue counters show forwarded and dropped packets per FC/queue; distinguish WRED drops from tail drops.
- Ingress policing drops appear at the SAP ingress meter; egress shaping drops appear at the queue or port scheduler.
- Telemetry (gNMI/MDT, SNMP) streams queue and drop counters; thresholds and alarms catch SLA breaches.
- Nokia NSP/assurance correlates service and transport KPIs; OAM (ETH-CFM, Y.1731) measures frame loss, delay and jitter.
- Baseline before changing policy; a policy change can shift drops to another queue rather than remove them.
Worked example
Users report voice quality loss on a SAP. show qos sap-ingress shows EF forwarded but out-profile packets increasing, and meter drops at 20 Mbps. The meter is policing voice at 20 Mbps while the codec plus signalling bursts to 24 Mbps. The fix is to size the meter to the signed voice SLA or re-mark signalling to AF41, not to raise EF to line rate. Checking the egress queue confirms no WRED drops, so the fault is ingress policing.
Traps
- Reading aggregate port drops instead of per-queue/per-FC counters.
- Confusing WRED early drops with a hard policer drop.
- Forgetting that egress drops are often caused by an upstream shaper or port scheduler.
- Not checking both directions; voice is bidirectional.
- Changing scheduler priority without checking the port scheduler's available bandwidth.