PDA

View Full Version : QoS in Subinterface


m.stone
05-24-2010, 10:44 PM
Hi, all

One of my customer is connected through a dot1 Q subinterface on our router . Is there is any way to implement this SLA for his service?

policy-map WAN-QoS
class http
priority 3000
set dscp af11
class Mail
bandwidth 2000
set dscp af41
class class-default
fair-queue

Regards

David.C
05-24-2010, 11:18 PM
Hi,

Check out the below link for QoS configuration in sub interface.

https://www.cisco.com/en/US/tech/tk543/tk545/technologies_tech_note09186a0080114326.shtml

Hope to help.

m.stone
05-31-2010, 03:36 AM
Hi,

Check out the below link for QoS configuration in sub interface.

https://www.cisco.com/en/US/tech/tk543/tk545/technologies_tech_note09186a0080114326.shtml

Hope to help.

Thanks

The below config. works with me on sub-interface , but what's the difference between this two ? Whether in MQC , if Im not using any bandwith on Mail and class-dfeault whether i will be able to shoot more in HTTP ?

Case1 : MQC
policy-map WAN-QoS
class http
shape average 3072000
set dscp af41
class Mail
shape average 1024000
set dscp af41
class class-default
fair-queue

Case2: CBWFQ
policy-map WAN-QoS
class http
priority 3000
set dscp af11
class Mail
bandwidth 2000
set dscp af41
class class-default
fair-queue

David.C
05-31-2010, 10:29 PM
The two cases whcih you have mentioned will perform different operations as below,

Case 1:
- HTTP traffic will be shaped to 3 Mbps and will be marked as AF41. HTTP traffic more than 3 Mbps will be buffered and treated with FIFO queueing mechanism by default and if the queue is full, packets will be dropped.
- Mail traffic will be shaped to 1 Mbps and will be marked as AF41. <ail traffic more than 1 Mbps will be buffered and treated with FIFO queueing mechanism by default and if the queue is full, packets will be dropped.
- Rest all traffic will get WFQ treatment.

Case 2:
- HTTP traffic upto 3 Mbps will be prioritised and will be sent out of the interface first in case there is a congestion.
- Mail traffic will have 2 Mbps even if the interface is congested.

Ay queuing will get kicked in only when it realise that the interface is congested (the hardware queue will be full and the packets will be in software queue). Basically CBWFQ/LLQ (Queuing mechanism) cannot be applied directly on subinterfaces. So you need to configure HQoS where you will first configure a parent policy to shape the traffic as per the CIR of that subinterface and then apply the Queuing policy as child policy.

For example, if you have a subinterface with CIR as 10Mbps and you need HTTP traffic to have 3 Mbps and prioritised always during the time of issue and Mail traffic to have 2 Mbps, you need to configure HQoS as below,

policy-map CHILD
class HTTP
priority 3000
set dscp af11
class Mail
bandwidth 2000
set dscp af41
class class-default
fair-queue

policy-map PARENT
class class-default
shape average 10000000
service-policy CHILD

The above configuration will shape all outbound traffic on this interface to 10 Mbps. Any traffic more than 10 Mbps will be buffered in shaping Queue. By default teh shaping Queue will be FIFO. Now that we have applied a CHILD policy, we made the FIFO queue to LLQ.