Can Apply QoS policy map to MPLS PE router mpls interface?

[Herman2018]

Hi, good day! can anyone pls advise whether we can apply QoS policy map to PE router MPLS interface or not? If can't, how to implement QoS on PE router to control the user traffics? Thanks in advance!

[marce1000]

 - Here is an example  creating  a policy that marks voice traffic with a DSCP value of EF (Expedited Forwarding), and limits the amount of bandwidth that non-voice traffic can consume to 50% of the available bandwidth :

           ip access-list extended VOICE
           permit udp any any range 16384 32767
          !
          class-map VOICE
          match access-group name VOICE
          !
          policy-map QOS
          class VOICE
         set dscp ef
         class class-default
         shape average 50%
         ! 
         !  Apply the QoS policy to the MPLS interface:
         !
         interface GigabitEthernet 0/0/0
         service-policy output QOS

                   Adapt and or change according to your needs, 

 M.

[Herman2018]

Thanks Marce for your advice!  Can QoS  traffic shaping applied on MPLS PE egress interface (leading to P routers) control the traffics based on dscp values? I check MPLS qos , the docs mention EXP bit, what is it? Look forward to your advice, thanks. 

[marce1000]

 - EXP (Experimental) bit is a 3-bit field in the MPLS header that can be used for Quality of Service (QoS) purposes, such as traffic prioritization and shaping. The EXP field allows up to 8 different levels of priority to be assigned to MPLS packets. In Cisco's implementation of MPLS, the EXP field is also referred to as the Class of Service (CoS) field. When a packet enters a Cisco MPLS network, the EXP field can be mapped to the Differentiated Services Code Point (DSCP) field in the IP header, which can then be used by QoS mechanisms such as traffic shaping, policing, and queuing to provide differentiated treatment to different types of traffic. When configuring traffic shaping on a Cisco MPLS PE router, you can specify the maximum rate at which traffic is allowed to be sent out of an interface and also specify the priority of traffic based on the EXP (CoS) field value. This allows you to control the flow of traffic and ensure that higher-priority traffic is given precedence over lower-priority traffic.

  M.

[M02@rt37]

Hello @Herman2018,

In MPLS networks, the EXP bits are used to mark packets with a priority value, which is used by the MPLS network to determine the forwarding treatment of the packet. The EXP bits are located in the MPLS header and are equivalent to the IP DSCP values.

When packets are sent across the MPLS network, the MPLS switches (or P routers) can be configured to perform QoS based on the EXP values. The MPLS switches can map the EXP values to different QoS policies, such as priority queuing, bandwidth allocation, or traffic shaping.

To summarize, you can use DSCP values to mark and classify traffic at the edge of the network, and then use EXP values to provide QoS treatment within the MPLS network. This allows you to maintain end-to-end QoS treatment for your traffic across the MPLS network.

[Herman2018]

Thanks M02@rt37 @marce1000 ! we can only apply QoS to egress interfaces (to P routers) of PE router because direct user VRF is configured on the VLAN interface on the PE router. Like this, can still classify the packets based on dscp values and service ports? the traffics are not encrypted , right? Please advise ,thanks a lot. 

[M02@rt37]

Hello @Herman2018,

QoS can be applied to both ingress and egress interfaces of a PE router. However, when applying QoS on a PE router that is serving multiple VRFs, it is generally recommended to apply QoS policies on egress interfaces towards the P routers to ensure that the QoS policies are applied uniformly across all VRFs.

When packets are classified based on DSCP values and service ports, the QoS policies will still work even if the traffic is not encrypted. QoS policies operate at the network layer and are independent of the data being carried by the packets.

However, if the traffic is encrypted, it may not be possible to classify the packets based on the contents of the packets, including DSCP values and service ports. In such cases, QoS policies may have to be based on the source or destination IP address, or other header fields that are not encrypted.

[Joseph W. Doherty]

What you can do with QoS would depend on the capabilities of the router, including whatever restrictions a MPLS interface might impose.

For example, a MPLS interface might preclude the router from looking deeper than going beyond the most external MPLS label's information (and since MPLS can stack MPLS labels, I would expect this to be a routine restriction [somewhat like, by default, Cisco routers cannot examine GRE contents beyond the GRE IP header]).

As other posters have noted, MPLS labels have an EXP field (3 bits, much like L2 COS or IP Precedence), which can be used for ToS.  Assuming that EXP is configured, you should be able to use it for QoS purposes, again, depending the the capabilities of the router.

Also, BTW, Cisco, for sometime now, when doing GRE, copies the original packet's ToS to the GRE header's ToS.  I recall (?) MPLS might do something similar with its labels, but if it does, it's limited to just 3 bits, which likely would be the first 3 bits of the IP ToS.