(Optional) Configuring an Flow Queue

You can configure scheduling parameters, traffic shaping for FQs in common 8-CoS mode based on network requirements.

Context

You can use a non-default FQ profile to configure scheduling parameters for FQs based on network requirements. Multiple flow WRED objects can be created for FQs to reference. You can set upper and lower drop thresholds and a drop probability for a created flow WRED profile. When the percentage of the actual length of a queue over the length of an FQ is less than the lower drop threshold, the system does not drop packets; when the percentage of the actual length of a queue over the length of an FQ is between the upper and lower drop thresholds, the system drops packets through the WRED mechanism (the longer the queue length, the higher the drop probability); when the percentage of the actual length of a queue over the length of an FQ is greater than the upper drop threshold, the system drops all subsequent packets.

The upper and lower drop thresholds for red packets can be set to the minimum; those for yellow packets can be greater; those for green packets can be set to the maximum.
In the actual configuration, the lower drop threshold is recommended to begin with 50% and be adjusted based on different colors of packets. 100% is recommended for the drop probability.
If no flow WRED object is set, the system adopts the default tail drop policy.

Procedure

Configure a flow WRED object.
1. Run system-view
  The system view is displayed.
2. Run flow-wred wred-name
  A flow WRED object is created and its view is displayed.
3. Run color { green | yellow | red } low-limit low-limit-percentage high-limit high-limit-percentage discard-percentage discard-percentage
  Upper and lower drop thresholds as well as a drop probability are configured.
4. Run queue-depth queue-depth-value
  A queue depth is set.
5. Run commit
  The configuration is committed.
6. Run quit
  Return to the system view.
Configure scheduling parameters for an FQ.
1. Run flow-queue flow-queue-name
  The FQ view is displayed.
2. Run queue cos-value { { pq | wfq weight weight-value | lpq } | { shaping { shaping-value | shaping-percentage shaping-percentage-value } [ pbs pbs-value ] | car { car-value | car-percentage car-percentage-value } [ pbs pbs-value ] } | flow-wred wred-name | low-latency | low-jitter } * or queue cos-value cir { { cir-value [ cbs cbs-value ] cir-schedule pq pir pir-value } | { cir-percentage cir-percentage-value [ cbs cbs-value ] cir-schedule pq pir pir-percentage pir-percentage-value } } [ pbs pbs-value ] pir-schedule { pq | wfq weight weight-value | lpq } [ flow-wred wred-name ]
  The scheduling parameters and policy of the FQ are modified.
Configure share shaping.
1. Run share-shaping [ shap-id ] { af1 | af2 | af3 | af4 | be | cs6 | cs7 | ef } ^* [ pq | wfq weight weight-value | lpq ] shaping-value [ pbs pbs-value ]
  Share shaping is configured for multiple FQs.After queues configured with share shaping are shaped, the queues are scheduled together with other user queues.
2. Run commit
  The configuration is committed.
3. Run quit
  Return to the system view.
Enable the low-latency function for FQs in PQ scheduling mode.
1. Run qos flow-queue low-latency enable
  The low-latency function is enabled for FQs in PQ scheduling mode.
Configure the PQ scheduling priority, the burst size.
1. Run slot slot-id
  The slot view is displayed
2. Run qos pq-scheduler priority { high | low } { inbound | outbound }
  The PQ scheduling priority on the board is changed.
3. Run qos cos { be | af1 | af2 | af3 | af4 | ef | cs6 | cs7 } burst-size buffer-size-value
  The burst size of a module is configured.
4. Run qos cos all pack-size pack-size-value
  The packet block threshold size is set.
5. Run commit
  The configuration is committed.
6. Run quit
  Return to the system view.