Feature Requirements |
Series |
Models |
|---|---|---|
1, The slot VE SR LPU1T board does not support upstream HQoS. 2, The downstream slot VE supports only HQoS in hardware loopback mode. 3, The global VE SR LPU1T board does not support upstream HQoS. 4, The downstream global VE supports HQoS in software/hardware loopback mode. If the software loopback mode is configured, downstream HQoS is looped back to upstream for implementation. For example, in L2 access L3 scenarios, HQoS is looped back from L2VE downstream to L3VE upstream for implementation. If HQoS is configured for L3VE upstream, HQoS does not take effect for L2VE downstream. 5, Both the software and hardware loopback of the LPUF200C/LPUF200D/LPUF240F use the upstream TM resources of the board. You can run the "display QoS resource user-queue slot <slot-id> {inbound | outbound}" command to query the resource usage. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
When member interfaces of a trunk reside on different TM/NP modules and traffic is not evenly balanced among trunk member interfaces using hash or traffic is not forwarded in load balancing mode, the rate limit on a single TM/NP module may be smaller than the configured value after bandwidth allocation. When load balancing among trunk member interfaces is uneven, packet loss may occur even if the traffic rate does not exceed the rate limit. Please do not configure inter-subcard member interfaces for a trunk. |
NetEngine 8000 F |
NetEngine 8000 F2A |
The CIR of a port queue and share-shaping are mutually exclusive. If the CIR is configured for a port queue on an interface, share-shaping cannot be configured on the interface. If share-shaping is configured on an interface, the CIR cannot be configured for a port queue on the interface. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The CS7, CS6, EF, AF4, AF3, AF2, AF1, and BE queues are serviced in descending order of priority. The low-latency feature of the three high-priority queues (CS7, CS6, and EF) are preferentially ensured. Low-latency services must be deployed in the three high-priority queues. If they are deployed in the other five low-priority queues, you must ensure that the three high-priority queues are not congested. Otherwise, the maximum latency of low-latency services in the five low-priority queues cannot be ensured. For example: In congestion scenarios, if the low-latency feature is configured for the BE queue not the CS7 queue, the average latency of the BE queue can be ensured, but the maximum latency cannot. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Upstream traffic on the board does not support SQ and can be scheduled only using HCAR. Compared with SQ, HCAR does not provide buffer queues and may encounter the following problems: 1. Token buckets are used to perform burst control. When the rate of burst traffic is higher than the token bucket setting, the traffic is simply discarded instead of being buffered. 2. When the traffic burst is large and the sum of weights of FQ services is large, WFQ scheduling is inaccurate. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
If the qos schedule-tree distribute-mode outbound command has been run on an Eth-Trunk interface and the Eth-Trunk has member interfaces, services will be interrupted during configuration rollback. After the configuration rollback is complete, services are automatically restored. When the scheduling tree split function is deleted, member interfaces in the trunk can be deleted only after they are removed from the trunk. As a result, services are affected. When the scheduling tree split function is rolled back, this operation can be implemented only after the member interfaces are removed from the trunk. Therefore, services are affected. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Port queues configured with the CIR and port queues configured with low delay preempt bandwidth. If the CIR is configured for a port queue, the port queue preempts bandwidth with port queues configured with low delay on the same interface. The priorities of port queues in bandwidth preemption are the same as the original priorities of port queues. The CIR and low delay cannot be both configured for a port queue. Please plan services properly. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The CIR can be configured for a maximum of two FQs in each SQ and can be configured for only two PQs on each interface. The CIR function cannot be configured for low-priority queues, and the bandwidth of low-priority queues cannot be guaranteed. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Flow queues configured with the CIR and flow queues configured with low delay preempt bandwidth. If the CIR is configured for a flow queue, the flow queue preempts bandwidth with flow queues configured with low delay in the same user queue. The priorities of flow queues in bandwidth preemption are the same as the original priorities of flow queues. The CIR and low delay cannot be both configured for a flow queue. Please plan services properly. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The CIR function of FQ and port-queue is supported only in the downstream direction. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The CIR of a flow queue and share-shaping are mutually exclusive. If the CIR is configured for a flow queue in a user queue, share-shaping cannot be configured in the user queue. If share-shaping is configured in the user queue, the CIR cannot be configured for the flow queue in the user queue. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Channelized sub-interfaces are mutually exclusive with IP hard pipes on the same main interface. Please plan services properly. |
NetEngine 8000 F |
NetEngine 8000 F1A |
The "qos-profile" and "user-group-queue" commands cannot be configured on a channelized sub-interface. Please plan services properly. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Configuring rate limiting on forwarding interfaces used by the hard pipe on the network side is not recommended. The QoS configuration on the main interface or sub-interface used for hard pipe forwarding (interface CAR for incoming and outgoing traffic, and interface ACL for incoming and outgoing traffic) also applies to hard pipe traffic. For example, if CAR is configured for incoming traffic on the P node, CAR applies to the hard pipe traffic, which may cause packet loss of the hard pipe. Do not deploy QoS car or traffic-policy which contains car/user-queue/deny actions on the network-side interface configured with hard pipes. |
NetEngine 8000 F |
NetEngine 8000 F1A |
As HCAR does not have the buffer capability, inbound HCAR causes inaccurate WFQ scheduling when a large traffic burst occurs and the sum of the FQ service weights among users is large (weight values). Configure a large HCAR bucket depth. You can use the PBS configuration tool provided to calculate a proper PBS value. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
After a channelized sub-interface is configured on a main interface, "sub-port-queue" cannot be configured on the main interface and all its sub-interfaces. Please plan services properly. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The device support inbound HCAR. As HCAR does not have the buffer capability, inbound HCAR causes inaccurate WFQ scheduling when a large traffic burst occurs and the sum of the FQ service weight values among users is large. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Sufficient CIR is a necessary condition for ensuring low delay. The default CIR of an interface is 10 Mbit/s. To ensure that the rate of low-delay port queues is greater than 10 Mbit/s, run the "qos default user-queue" command to change the default CIR of the interface to a value that is not lower than the traffic bandwidth of the low-delay port-queues.If the bandwidth of low-delay services is greater than the CIR, low delay cannot be guaranteed for services. Run the "qos default user-queue" command to adjust the default SQ CIR on an interface. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
When interface-based HQoS is configured on a GRE tunnel interface, if the incoming packets on the GRE tunnel are sent from multiple interface boards or multiple TM modules, the HQoS rate limit multiplies based on the number of TM modules. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
On a global VE interface, the SQ configured for outbound flows does not take effect. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Consecutive VLANs and QoS profiles are configured on an interface. If two or more inconsecutive VLANs are deleted and committed at the same time, service flapping and statistics data may be lost. If you delete consecutive VLANs and QoS profile from an interface, delete one or consecutive VLANs at a time and then submit the configuration. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
If user-queue is configured in a traffic policy applied to a channelized sub-interface, the user-queue bandwidth is restricted by the remaining unchannelized bandwidth on the main interface, not the channelized sub-interface bandwidth. Therefore, you are not advised to configure user-queue in a traffic policy applied to a channelized sub-interface. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The sum of the channelized sub-interface bandwidths configured on the same physical interface cannot exceed 98% of the physical interface bandwidth minus the maximum reservable bandwidth of the TE tunnel. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Channelized sub-interfaces can be configured only on VLAN-type dot1q sub-interfaces of GE, XGE, 10GE, 25GE, 40GE, 50GE, 100GE, 200GE, and 400GE physical interfaces, FlexE physical interfaces, as well as Eth-Trunk interfaces. You are advised to plan functions properly. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Channelized sub-interfaces cannot perform precise compensation based on services. If the maximum compensation value and precise shaper processing are used, the interface usage cannot reach 100%. You are advised to plan functions properly. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The channelized sub-interface does not support the user-group-queue command. You are advised to plan functions properly. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
If channelized sub-interfaces are configured on a main interface, IP hard pipe cannot be configured for these channelized sub-interfaces. After IP hard pipe is configured on an interface, channelized sub-interfaces cannot be configured on the interface. You are advised to plan functions properly. |
NetEngine 8000 F |
NetEngine 8000 F1A |
Eth-Trunk channelized sub-interfaces cannot be configured on Eth-Trunk member interfaces that reside on the same board but have different TM types. You are advised to plan functions properly. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
The Eth-Trunk channelized sub-interface function and Eth-Trunk scheduling tree splitting are mutually exclusive. You are advised to plan functions properly. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
If channelized sub-interfaces are configured on a main interface, the "sub-port-queue" and "user-group-queue" cannot be run on the main interface and its all sub-interfaces. You are advised to plan functions properly. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
If channelized sub-interfaces are configured on a main interface, the "qos default { sub-port-queue | user-group-queue }" command cannot be run on the main interface. You are advised to plan functions properly. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Configuring channelized sub-interfaces on an Eth-Trunk interface is mutually exclusive with running the "port shaping" command on member interfaces of the Eth-Trunk interface. You are advised to plan functions properly. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
Channelized sub-interfaces do not support the "qos-profile," "user-queue," or "user-queue shaping ldp-traffic outbound" command. You are advised to plan functions properly. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
If a channelized sub-interface is configured on an Eth-Trunk interface, member interfaces of the Eth-Trunk interface do not support the qos default user-queue command. You are advised to properly plan the function. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
When Eth-Trunk member interfaces reside on different TMs, the bandwidth configured for channelization is delivered based on the weight and equal ratio of the same TM by default, and the bandwidth is not doubled. When the hash algorithm is uneven, the bandwidth and delay for channelization cannot be reached, and packet loss occurs. You are advised to properly plan the function. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
If the "port shaping" command is not run on an Eth-Trunk interface, the total bandwidth of channelized sub-interfaces cannot exceed 98% of the Eth-Trunk interface's bandwidth. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
By default, Eth-Trunk channelized sub-interface splits the channelized bandwidth based on the TM weight ratio. When a member interface is added to or removed from an Eth-Trunk, the weight ratio of the TM to which each member interface belongs changes, and therefore the channelized bandwidth is updated. During this period, the channelized traffic cannot reach the limited bandwidth or delay, packet loss occurs. You are advised to properly plan the function. |
NetEngine 8000 F |
NetEngine 8000 F2A/NetEngine 8000 F1A |
When traffic of an Eth-Trunk channelized sub-interface is unevenly balanced, its bandwidth cannot be ensured. After the <b>mode channel bandwidth maximize</b> command is run on the Eth-Trunk channelized sub-interface to enable the extension mode, the channelized bandwidth can be increased. If the involved interface is not congested, the delay can be ensured. However, if the involved interface is congested, the delay cannot be ensured. Adjust the channelized bandwidth to a value greater than the actual one to solve the congestion problem. |
NetEngine 8000 F |
NetEngine 8000 F2A |