14. Quality of service

14.1. Overview

The SR Linux supports policies for assigning traffic to forwarding classes or remarking traffic at egress before it leaves the router. DSCP classifier policies map incoming packets to the appropriate forwarding classes, and DSCP rewrite-rule policies mark outgoing packets with an appropriate DSCP value based on the forwarding class.

Note:

Differences in how QoS functions on 7250 IXR systems compared with 7220 IXR-D2 and D3 systems are noted where applicable. Functionality is not supported on the 7220-D1.

14.1.1. How QoS works for transit traffic with a 7250 IXR chassis

The following describes how transit unicast packets flow through the SR Linux with a 7250 IXR chassis-based system:

  1. Packets are received on a subinterface and determined to be transit.
  2. Each received packet is classified as belonging to one of eight forwarding classes. The classification depends on the packet type and subinterface configuration, as follows:
    1. If the received packet is IPv4 or IPv6, and there is no DSCP classifier policy applied to the input subinterface, the system default DSCP classifier policy (with the reserved name default) is used. See Table 10 for the values assigned to the system default DSCP classifier policy.
    2. If the received packet is IPv4 or IPv6, and there is a non-default DSCP classifier policy applied to the input subinterface, the classifier policy tries to match the 6-bit DSCP value in the IP header to one of its entries.
      Each policy entry associates one or more DSCP values with forwarding class 0 to 7. If there is no entry matching the received DSCP value, the assigned forwarding class is 0.
  3. A forwarding lookup on the relevant packet header determines its egress line card, egress port, and egress queue. The egress queue is based on the forwarding class of the packet. If the packet is classified as forwarding class 0 (fc0), it is associated with egress queue0. If the packet is classified as fc1 then it is associated with egress queue1, and so on.
  4. The packet is added to a virtual output queue (VoQ) associated with the egress queue; when the packet reaches the head of the VoQ, it is transmitted to the egress queue. If the selected VoQ is at maximum depth, the packet is dropped.
  5. At the egress IMM, there are two egress queues (EGQs) per port, one for unicast traffic and one for multicast traffic. There is a WRR relationship between the two EGQs, with the unicast EGQ having a non-configurable weight of 9 and the multicast EGQ having a non-configurable weight of 1.
  6. The bandwidth available to the unicast EGQ is divided between the different forwarding classes based on the peak information rate (PIR) and scheduling configuration (strict priority or weighted round robin) of each FC. By default, every FC is set up as strict priority.
  7. Strict priority FCs are served first (in descending order of FC) and then the WRR FCs (by weight). limiting each forwarding class to its PIR (configured as a percentage of the egress port bandwidth); by default, the PIR of each forwarding class is 100%.
  8. The Class of Service (CoS) field in the outgoing IPv4/IPv6 packet header is written or modified as follows:
    1. The DSCP value is left unchanged if no DSCP rewrite-rule policy is applied to the output subinterface.
    2. The DSCP value is changed to value X if there is a DSCP rewrite-rule policy applied to the output subinterface, and there is a configured entry in that policy that matches the forwarding class of the packet and specifies DSCP value X.
    3. The DSCP value is changed to 0 if there is a DSCP rewrite-rule policy applied to the output subinterface, but there is no configured entry in that policy that matches the forwarding class of the packet.
      Table 10:  Values assigned to the system default DSCP classifier policy  

      DSCP values

      Included DSCP names

      Forwarding class

      0 to 7

      CS0/BE

      fc0

      8 to 15

      CS1, AF11 to 13

      fc1

      16 to 23

      CS2, AF21 to 23

      fc2

      24 to 31

      CS3, AF31 to 33

      fc3

      32 to 39

      CS4, AF41 to 43

      fc4

      40 to 47

      CS5, EF

      fc5

      48 to 55

      CS6/NC1

      fc6

      56 to 63

      CS7/NC2

      fc7

Note:

On ingress VLAN ports, VLAN priority is ignored. If the egress interface is a VLAN-enabled interface, the Priority Code Point (PCP) bits are reset to 0. ECN bits are preserved

14.1.2. How QoS works for router-originated traffic with a 7250 IXR chassis

For IPv4/IPv6 packets originated on a 7250 IXR chassis-based system, QoS works as follows:

  1. An application on the SR Linux CPM generates an IPv4/IPv6 packet or Ethernet frame to send to another system.
  2. A forwarding lookup on the relevant packet header determines its egress line card, egress port, and egress queue. The egress queue is based on the forwarding class of the packet, shown in Section 14.1.7, Table 12. If the packet is classified as forwarding class 0 (fc0), it is associated with egress queue0. If the packet is classified as fc1, it is associated with egress queue1, and so on.
  3. The packet is sent to the egress line card and added to a virtual output queue (VoQ) associated with the egress queue; when the packet reaches the head of the VoQ, it is transmitted to the egress queue. If the selected VoQ is at maximum depth, the packet is dropped
  4. Egress queues are scheduled directly to the output port.

The DSCP field in the outgoing IPv4/IPv6 packet header is written or modified as listed in Section 14.1.7, Table 12. For router-originated traffic, the values in this table override any rewrite-rule policy applied to the output subinterface.

14.1.3. How QoS works for router-terminated traffic with a 7250 IXR chassis

For IPv4/IPv6 packets terminated on a 7250 IXR chassis-based system, QoS works as follows:

  1. Packets are received on a subinterface and determined to need extraction towards the CPM.
  2. If the ingress port is 1-20 or 25-28 then the selected VOQ resides on core0. All other ingress ports reside on core1.
  3. The VOQ selected is based on the first matching PMF rule.
  4. Each VOQ has a maximum depth (MBS) of 20 MB. The packet is queued unless it exceeds this MBS; otherwise the packet is dropped.
  5. The following PIR rate limits apply to CPM VOQs:
    1. Limited to 1 Mbps: VOQs associated with packets hitting IPv4 blackhole-reject routes, packets hitting IPv6 blackhole-reject routes, destination unknown traffic, traffic with header errors and traffic triggering ICMP redirects
    2. Limited to 2 Mbps: VOQs associated with ARP traffic, traffic that exceeds the outgoing interface MTU, traffic with expiring TTL, and L2 broadcast traffic
    3. Limited to 4 Mbps: VOQ associated with LLDP traffic
    4. Limited to 7 Mbps: VOQs associated ACL log traffic and capture-filter traffic

14.1.4. How QoS works for transit traffic with a 7220 IXR-D2 and D3

The following describes how transit packets flow through on a 7220 IXR-D2 and D3 system:

  1. Packets are received on a subinterface.
  2. Each received packet is classified as belonging to one of eight forwarding classes. The classification depends on the traffic flow and packet type as shown in Table 11.
  3. A forwarding lookup determines the egress port.
  4. The packet is directed to one of 16 egress queues associated with the output port.
  5. The packet is dropped if the on-chip buffer memory is exhausted; otherwise it is stored in the memory (32 MB).
  6. The unicast and multicast queue for the same forwarding class are known as a queue pair. There are eight possible queue pairs for an egress port .
    1. Each pair is associated with a top-level scheduler node.
    2. Each scheduler node is served as strict priority or WRR, and if it is served as WRR, it also has an associated weight. The discipline and WRR weight of each scheduler node comes from the configuration of the unicast queue in each queue pair.
  7. All unicast queues and scheduler nodes are setup as Strict Priority (SP) by default. Scheduler nodes can be changed from SP to WRR. Each WRR node has a weight in the range 1-127. The scheduling loop serves the strict priority nodes first (in descending order of FC) and then the WRR nodes (by weight).
  8. Each unicast queue can be configured with a PIR. PIR is defined as a percentage of egress port bandwidth. The default of each queue is 100%.
  9. The ECN bits are transmitted unchanged (as received at ingress) regardless of whether or not the DSCP is rewritten.
Note:

There is no separate DSCP classifier for IPv4 and IPv6 with the 7220 IXR-D2 and D3. There is one policy applied for both IPv4 and IPv6 packets.

Table 11:  Forwarding class assignment rules (7220 IXR-D2 and D3) 

Traffic flow

Packet type

Forwarding class assignment

bridged -> bridged switched at L2

IP

  1. If the configuration of the ingress bridged subinterface refers to a DSCP classifier policy and the qos_mgr is up
  2. Then determine the FC from that mapping table (even if no entry exists for the DSCP value)
  3. Otherwise, determine the FC from the default DSCP classifier policy

bridged -> bridged switched at L2

non-IP

  1. FC is set to default-fc value of the ingress bridged subinterface (default fc0)

bridged -> routed through IRB

IP

  1. If the configuration of the IRB subinterface refers to a DSCP classifier policy and the qos_mgr is up
  2. Then determine the FC from that mapping table (even if no entry exists for the DSCP value)
  3. Otherwise, determine the FC from the default DSCP classifier policy

routed -> routed

IP

  1. If the configuration of the ingress routed subinterface refers to a DSCP classifier policy and the qos_mgr is up
  2. Then determine the FC from that mapping table (even if no entry exists for the DSCP value)
  3. Otherwise, determine the FC from the default DSCP classifier policy

14.1.5. How QoS works for router-originated traffic with a 7220 IXR-D2 and D3

For IPv4/IPv6 packets originated on a 7220 IXR-D2 and D3 system, QoS works as follows:

  1. An application on the SR Linux CPM generates an IPv4/IPv6 packet or Ethernet frame to send to another system.
  2. The system assigns a DSCP to the self-generated packet based on its protocol and the hardcoded mapping shown in Section 14.1.7, Table 12. Except for ICMP and ICMPv6 echo-request packets, the DSCP values cannot be overridden.
  3. The system looks up the DSCP in the system-default dscp-classifier policy to determine the forwarding-class.
  4. A forwarding lookup determines the egress port.
  5. The packet is sent to the appropriate egress queue for its forwarding class and packet type. The decision to drop or enqueue the packet in the egress queue and the scheduling of the egress queue follows the same steps as transit traffic
  6. The DSCP field in the IPv4/IPv6 header is always written based on the hardcoded mapping shown in Section 14.1.7, Table 12. If the packet also matches a dscp-policy rewrite-rule applied to the output subinterface, then the mapping is ignored for the originated packet.

14.1.6. How QoS works for router-terminated traffic with a 7220 IXR-D2 and D3

For packets terminated on a 7220 IXR-D2 and D3 based system, a packet is received on a subinterface and determined to need extraction towards the CPM. The packet is directed to one of the 48 queues associated with the CPM as a destination 'physical port' based on its protocol or type.

The 7220 IXR-D2 and D3 have a 4MB CPU queue buffer that is shared across all CPU queues for CPU-terminated traffic.

14.1.7. Self-generated traffic classification and marking

Self-generated traffic is classified and marked based on the hard-coded mapping rules

Table 12:  Default forwarding class and DCSP marking for router-originated traffic  

Protocol / message type

Forwarding class

DSCP marking

IPv4 ARP request/reply

6

N/A

ICMPv4 including echo-request1, echo- reply2, dest-unreachable, redirect, time-exceeded, parameter-problem

0

0

ICMPv4 echo-request with ToS/DSCP override = X

(lookup X in system-default DSCP classifier)

X

ICMPv4 echo-reply to echo-request with non-zero DSCP X

(lookup X in system-default DSCP classifier)

X

UDP traceroute

0

0

IPv6 neighbor solicitation

6

48 (CS6/NC1)

IPv6 neighbor advertisement

6

48 (CS6/NC1)

All other ICMPv6 including dest unreachable, packet-too-big, time- exceeded, parameter-problem, echo-request, echo-reply, router-solicitation, redirect

0

0

ICMPv6 echo-request with DSCP override = X

(lookup X in system-default DSCP classifier)

X

ICMPv6 echo-reply to echo-request with non-zero DSCP X

(lookup X in system-default DSCP classifier)

X

BFD

6

48 (CS6/NC1)

BGP

6

48 (CS6/NC1)

DNS query

4

34 (AF41)

FTP/TFTP

4

34 (AF41)

gNMI

4

34 (AF41)

JSON RPC

4

34 (AF41)

LLDP

N/A

N/A

NTP

4

34 (AF41)

sFlow

0

0

SNMP

4

34 (AF41)

SSH

4

34 (AF41)

Syslog

4

34 (AF41)

TACACS+

4

34 (AF41)

1 Echo-request generated by a ping command with no DSCP parameter specified

2 Echo-reply to an echo-request packet with DSCP=0

14.2. Configuring QoS

QoS configuration on the SR Linux consists of creating classifier policies for incoming packets, rewrite-rule policies for outgoing packets, and applying the policies to subinterfaces in the inbound or outbound direction.

14.2.1. Configuring QoS policies

14.2.1.1. Classifier policies

When a classifier policy is applied to a subinterface, the policy attempts to match the 6-bit DSCP value in the IP header of incoming packets to one of its entries. If there is a match, the incoming packet is assigned to the specified forwarding class; otherwise, the assigned forwarding class is 0.

Example:

The following example specifies a classifier policy:

--{ candidate shared default }--[  ]--
# info qos classifiers
 qos {
        classifiers {
            dscp-policy p1 {
                dscp 0 {
                    forwarding-class fc7
                }
                dscp 8 {
                    forwarding-class fc6
                }
            }
        }
    }

14.2.1.2. Rewrite-rule policies

When a rewrite-rule policy is applied to a subinterface, the policy attempts to match the forwarding class of outbound packets to one of its entries. If there is a match, the DSCP value of the outbound packet is changed to the value specified by the policy. If the forwarding class does not match the rewrite-rule policy, the DSCP value is changed to 0.

Example:

The following example specifies a rewrite-rule policy:

--{ candidate shared default }--[  ]--
# info qos rewrite-rules
 qos {
        rewrite-rules {
            dscp-policy p2 {
                map fc0 {
                    dscp 1
                }
            }
        }
    }

14.2.2. Applying QoS policies to subinterfaces

14.2.2.1. Applying a classifier policy to input traffic

If you apply a classifier policy to input traffic on a subinterface, incoming packets are evaluated against the policy, and matching packets are assigned to the forwarding class specified by the policy. If no classifier policy is applied to the subinterface, the system default DSCP classifier (with the reserved name default) is used.

Note:

On 7220 IXR-D2 and D3 systems, separate classifier policies for IPv4 and IPv6 traffic are not supported, but you can apply a common policy that applies to both IPv4 and IPv6 traffic.

The following example applies a classifier policy to inbound IPv6 traffic on a subinterface with a 7250 IXR chassis:

Example (7250 IXR):

--{ candidate shared default }--[  ]--
# info interface ethernet-1/1
    interface ethernet-1/1 {
        subinterface 1 {
            qos {
                input {
                    classifiers {
                        ipv6-dscp p1
                    }
                }
            }
        }
    }

The following example applies a classifier policy to inbound traffic on a subinterface with a 7220 IXR-D2 and D3:

Example (7220 IXR-D2 and D3):

--{ candidate shared default }--[  ]--
# info interface ethernet-1/1
    interface ethernet-1/1 {
        subinterface 1 {
            qos {
                input {
                    classifiers {
                        dscp p1
                    }
                }
            }
        }
    }

14.2.2.2. Applying a rewrite-rule policy to output traffic

When a rewrite-rule policy is applied to output traffic on a subinterface, outbound packets are evaluated against the policy. For packets matching the policy, the DSCP value is changed to the value specified by the policy. If no rewrite-rule policy is applied to the subinterface, the DSCP value is left unchanged.

Note:

The DSCP rewrite policy that applies to ingress IPv4 and IPv6 packets on the subinterface is not supported on 7220 IXR-D2 and D3 systems.

The following example applies a rewrite-rule policy to outbound IPv4 traffic on a subinterface with a 7250 IXR chassis:

Example (7250 IXR):

--{ candidate shared default }--[  ]--
# info interface ethernet-1/1
    interface ethernet-1/1 {
        subinterface 1 {
            qos {
                output {
                    rewrite-rules {
                        ipv4-dscp p2
                    }
                }
            }
        }
    }

The following example applies a rewrite-rule policy to outbound traffic on a subinterface with a 7220 IXR-D2 and D3:

Example (7220 IXR-D2 and D3):

--{ candidate shared default }--[  ]--
# info interface ethernet-1/1
    interface ethernet-1/1 {
        subinterface 1 {
            qos {
                output {
                    rewrite-rules {
                        dscp p2
                    }
                }
            }
        }
    }

14.2.3. Configuring output queue scheduling

You can configure the queues for outbound traffic on an interface as either strict priority (SP), weighted round robin (WRR), or mixed mode (SP + WRR). Queues configured as SP are served in priority order from highest forwarding class to lowest before any of the WRR queues.

After the SP queues are served, the WRR queues use the remaining bandwidth according to their configured weights. By default, the output queues on each interface have a weight of 1.

Examples:

The following example configures a the unicast output queue on an interface to be strict priority. This queue is served prior to those configured as WRR. Note that when strict priority is set to false, the queue is configured as WRR. When strict priority is set to true, any configured weight is ignored.

--{ candidate shared default }--[  ]--
# info interface ethernet-1/1
    interface ethernet-1/1 {
        qos {
            output {
                unicast-queue 0 {
                    queue-parameters {
                        strict-priority true
                    }
                }
            }
        }
    }

The following example configures weight for the unicast output queue on an interface. After the strict priority queues are served, the other queues use the remaining bandwidth according to the configured weights. Queues that are not configured with a specific weight have a weight of 1.

--{ candidate shared default }--[  ]--
# info interface ethernet-1/1
    interface ethernet-1/1 {
        qos {
            output {
                unicast-queue 0 {
                    queue-parameters {
                        weight 20
                    }
                }
            }
        }
    }

The following example sets the maximum percentage of port bandwidth that is available to traffic in the unicast output queue. By default traffic in the unicast output queue can use up to 100% of port bandwidth.

--{ candidate shared default }--[  ]--
# info interface ethernet-1/1
    interface ethernet-1/1 {
        qos {
            output {
                unicast-queue 0 {
                    queue-parameters {
                        peak-rate-percent 75
                    }
                }
            }
        }
    }

14.3. Viewing buffer utilization

Buffer utilization differs between the 7250 IXR and the 7220 IXR-D2 and D3. The 7250 IXR utilizes high bandwidth memory (HBM) and VOQ, while the 7220 IXR-D2 and D3 do not. See Table 13.

Table 13:  Buffer utilization 

Hardware

Buffer memory

7250 IXR

  1. SRAM size = 32MB
  2. DRAM (HBM) size =8 GB
  3. Per VOQ limit = 256MB

7220 IXR-D2 and D3

  1. Total Buffer size = 32MB
  2. Reserved Buffer size = 4MB

The following example shows overall buffer usage. Depending on the hardware deployed, the output will vary.

The following shows output for a 7250 IXR:

# info from state platform linecard 1 forwarding-complex 0 buffer-memory
    platform {
        linecard 1 {
            forwarding-complex 0 {
                buffer-memory {
                    sram {
                        used 15808512 >> in bytes 
                        free 17745920  >> in bytes
                    }
                    dram {
                        used 48   >>> it is in % of BDB
                    }
                }
            }
        }
    }

The following shows output for a 7220 IXR-D2 and D3:

# info from state platform linecard 1 forwarding-complex 0 buffer-memory
    platform {
        linecard 1 {
            forwarding-complex 0 {
                buffer-memory {
                   used  2097152
                   free  27263246
                   reserved 4194034
                }
            }
        }
    }

14.4. Displaying QoS statistics

Use the info from state command to display information about traffic subject to classifier and rewrite-rule policies.

Examples:

The following example shows information about traffic subject to a specified classifier policy:

--{ running }--[  ]--
# info from state qos classifiers dscp-policy default dscp 9 
qos {
        classifiers {
            dscp-policy default {
                dscp 9 {
                    forwarding-class fc1
                }
            }
        }
    }

The following example shows information about traffic subject to a specified rewrite-rule policy.

# info from state qos rewrite-rules dscp-policy default map fc3
    qos {
        rewrite-rules {
            dscp-policy default {
                map fc3 {
                    dscp 24
                }
            }
        }
    }

The following example shows information for a specified subinterface:

--{ candidate shared default}--[  ]--
# interface ethernet-1/2
--{ candidate shared }--[ interface ethernet-1/2 ]--
# subinterface 1
--{ candidate shared }--[ interface ethernet-1/2 subinterface 1 ]--
# info from state qos
    qos {
        input {
            classifiers {
                ipv4-dscp default
                ipv6-dscp default
                mpls-tc default
            }
        }
        output {
            rewrite-rules {
                ipv4-dscp 7
            }
        }
    }

To display traffic statistics for each output queue on an interface, use the show interface <id> queue-statistics command in running or candidate mode, or the interface port detail command in show mode.

Example:

The following example displays output queue statistics for an interface:

# show interface ethernet-1/1 queue-statistics
===================================================================================
Interface: ethernet-1/1
-----------------------------------------------------------------------------------
  Description     : <None>
  Oper state      : down
  Down reason     : lower-layer-down
  Last change     : 4d14h50m28s ago, No flaps since last clear
  Speed           : 100G
  Flow control    : Rx is disabled, Tx is not supported
  Loopback mode   : false
  MTU             : 9232
  VLAN tagging    : false
  Queues          : 8 output queues supported, 0 used since the last clear
  MAC address     : 68:AB:09:A2:71:B0
  Last stats clear: never
-----------------------------------------------------------------------------------
Queue Parameter for for ethernet-1/1
-----------------------------------------------------------------------------------
  Queue-id   Scheduling   Weight   PIR %    PIR (kbps) 
  0          SP           -        100     98994140.625
  1          SP           -        100     98994140.625
  2          SP           -        100     98994140.625
  3          SP           -        100     98994140.625
  4          SP           -        100     98994140.625
  5          SP           -        100     98994140.625
  6          SP           -        100     98994140.625
  7          SP           -        100     98994140.625
-----------------------------------------------------------------------------------
Queue statistics for interface ethernet-1/1, Queue 0 (fc0 traffic)
-----------------------------------------------------------------------------------
           Name            Fwd-Octets   Fwd-Pkts   Drop-Octets   Drop-Pkts
  Unicast Egress queue     0            0          0             0        
  VOQ 1                    0            0          0             0        
  VOQ 2                    0            0          0             0        
  VOQ 3                    0            0          0             0        
  VOQ 4                    0            0          0             0        
  Multicast Egress queue   0            0          0             0        
-----------------------------------------------------------------------------------
Queue statistics for interface ethernet-1/1, Queue 1 (fc1 traffic)
-----------------------------------------------------------------------------------
           Name            Fwd-Octets   Fwd-Pkts   Drop-Octets   Drop-Pkts
  Unicast Egress queue     0            0          0             0        
  VOQ 1                    0            0          0             0        
  VOQ 2                    0            0          0             0        
  VOQ 3                    0            0          0             0        
  VOQ 4                    0            0          0             0        
  Multicast Egress queue   0            0          0             0        
-----------------------------------------------------------------------------------
Queue statistics for interface ethernet-1/1, Queue 2 (fc2 traffic)
-----------------------------------------------------------------------------------
           Name            Fwd-Octets   Fwd-Pkts   Drop-Octets   Drop-Pkts
  Unicast Egress queue     0            0          0             0        
  VOQ 1                    0            0          0             0        
  VOQ 2                    0            0          0             0        
  VOQ 3                    0            0          0             0        
  VOQ 4                    0            0          0             0        
  Multicast Egress queue   0            0          0             0        
-----------------------------------------------------------------------------------
Queue statistics for interface ethernet-1/1, Queue 3 (fc3 traffic)
-----------------------------------------------------------------------------------
           Name            Fwd-Octets   Fwd-Pkts   Drop-Octets   Drop-Pkts
  Unicast Egress queue     0            0          0             0        
  VOQ 1                    0            0          0             0        
  VOQ 2                    0            0          0             0        
  VOQ 3                    0            0          0             0        
  VOQ 4                    0            0          0             0        
  Multicast Egress queue   0            0          0             0        
-----------------------------------------------------------------------------------
Queue statistics for interface ethernet-1/1, Queue 4 (fc4 traffic)
-----------------------------------------------------------------------------------
           Name            Fwd-Octets   Fwd-Pkts   Drop-Octets   Drop-Pkts
  Unicast Egress queue     0            0          0             0        
  VOQ 1                    0            0          0             0        
  VOQ 2                    0            0          0             0        
  VOQ 3                    0            0          0             0        
  VOQ 4                    0            0          0             0        
  Multicast Egress queue   0            0          0             0        
-----------------------------------------------------------------------------------
Queue statistics for interface ethernet-1/1, Queue 5 (fc5 traffic)
-----------------------------------------------------------------------------------
           Name            Fwd-Octets   Fwd-Pkts   Drop-Octets   Drop-Pkts
  Unicast Egress queue     0            0          0             0        
  VOQ 1                    0            0          0             0        
  VOQ 2                    0            0          0             0        
  VOQ 3                    0            0          0             0        
  VOQ 4                    0            0          0             0        
  Multicast Egress queue   0            0          0             0        
-----------------------------------------------------------------------------------
Queue statistics for interface ethernet-1/1, Queue 6 (fc6 traffic)
-----------------------------------------------------------------------------------
           Name            Fwd-Octets   Fwd-Pkts   Drop-Octets   Drop-Pkts
  Unicast Egress queue     0            0          0             0        
  VOQ 1                    0            0          0             0        
  VOQ 2                    0            0          0             0        
  VOQ 3                    0            0          0             0        
  VOQ 4                    0            0          0             0        
  Multicast Egress queue   0            0          0             0        
-----------------------------------------------------------------------------------
Queue statistics for interface ethernet-1/1, Queue 7 (fc7 traffic)
-----------------------------------------------------------------------------------
           Name            Fwd-Octets   Fwd-Pkts   Drop-Octets   Drop-Pkts
  Unicast Egress queue     0            0          0             0        
  VOQ 1                    0            0          0             0        
  VOQ 2                    0            0          0             0        
  VOQ 3                    0            0          0             0        
  VOQ 4                    0            0          0             0        
  Multicast Egress queue   0            0          0             0        
===================================================================================

14.4.1. Clearing QoS statistics

You can reset the queue statistics counters for an interface.

Example:

The following example resets statistics counters for a specified egress queue (multicast) on an interface:

--{ running }--[  ]--
# tools interface ethernet-1/1 statistics queue-statistics multicast-queue 1 clear

14.4.2. Displaying QoS profile resource usage

A QoS profile resource refers to the number of classifier and rewrite policies that are applied to interfaces on a line card. Each classifier or rewrite policy that is applied to an interface on a line card counts as one profile resource used.

For example, if you create classifier policy dscp1 and apply it to input IPv4 traffic on an interface, and apply the same dscp1 policy to input IPv6 traffic on a different interface on the same line card, it counts as two classifier profile resources used.

The SR Linux supports up to 15 classifier profile resources and up to 32 rewrite profile resources per line card. You can display the number of QoS profile resources in use for each line card.

Example:

The following example displays the number of used and free classifier and rewrite profile resources for a line card:

# info from state platform linecard 1 forwarding-complex 0 qos
    platform {
        linecard 1 {
            forwarding-complex 0 {
                qos {
                    resource classifier-profiles {
                        used 1
                        free 15
                    }
                    resource rewrite-profiles {
                        used 1
                        free 31
                    }
                }
            }
        }
    }