- Finding Feature Information
- Prerequisites for QoS Policies Aggregation
- Restrictions for QoS Policies Aggregation
- Information About QoS Policies Aggregation
QoS Policies Aggregation
The QoS: Policies Aggregation (QoS: Policies Aggregation) feature for the Cisco ASR 1000 Series Aggregation Services Routers supports Modular Quality of Service Command-Line Interface (MQC) configuration of default traffic classes in policy maps on different subinterfaces to be queued as a single, user-defined traffic class at the main interface policy map. It is most useful in QoS configurations where you have several subinterface policy maps on the same physical interface and you want identical treatment of the default traffic classes on those subinterfaces.
Beginning in Cisco IOS XE Release 2.6, the QoS: Policies Aggregation feature is enhanced to support queueing aggregation at the primary interface for other traffic classes, including Differentiated Services Code Point (DSCP) traffic classes such as the expedited forwarding (EF), Assured Forwarding 1 (AF1), and AF4 traffic classes. With this enhancement, any traffic classes from VLAN subinterfaces can share a common queue for that traffic class at the main interface policy map. Other enhancements include the ability to configure and show drop statistics that occur at the aggregate level for these classes.
- Finding Feature Information
- Prerequisites for QoS Policies Aggregation
- Restrictions for QoS Policies Aggregation
- Information About QoS Policies Aggregation
- How to Configure QoS Policies Aggregation
- How to Configure QoS Policies Aggregation MQC
- Configuration Examples for QoS Policies Aggregation
- Additional References
- Feature Information for QoS Policies Aggregation
Finding Feature Information
Your software release may not support all the features documented in this module. For the latest feature information and caveats, see the release notes for your platform and software release. To find information about the features documented in this module, and to see a list of the releases in which each feature is supported, see the Feature Information Table at the end of this document.
Use Cisco Feature Navigator to find information about platform support and Cisco software image support. To access Cisco Feature Navigator, go to www.cisco.com/go/cfn. An account on Cisco.com is not required.
Prerequisites for QoS Policies Aggregation
- This feature is configured using the Modular Quality of Service (QoS) Command-Line Interface (CLI) (MQC).
- All traffic over the main interface should come through one or more subinterfaces.
Restrictions for QoS Policies Aggregation
- Applies only when multiple subinterfaces with policy maps are attached to the same physical interface. This feature cannot be used to collectively classify default traffic classes or other traffic classes of policy maps on different physical interfaces.
- Certain traffic class configuration prior to Cisco IOS XE Release 2.6 at the subinterface policy-map and main-interface policy-map will have different behavior and queueing results. See the Understanding the QoS Policies Aggregation MQC and Differences Between the Original Feature and the MQC Support for Multiple Queue Aggregation.
Information About QoS Policies Aggregation
- Understanding Fragments in Class Definition Statements
- Understanding Fragments for Gigabit Etherchannel Bundles
- Understanding the QoS Policies Aggregation MQC
Understanding Fragments in Class Definition Statements
QoS: Policies Aggregation introduces the idea of fragments in class definition statements. A default traffic class definition statement can be marked as a fragment within a policy map. Other policy maps on the same interface can also define their default traffic class statements as fragments, if desired. A separate policy map can then be created with a service fragment class definition statement that will be used to apply QoS to all of the fragments as a single group.
The figure below provides an example of one physical interface with three attached policy maps that is not using fragments. Note that each policy map has a default traffic class that can only classify traffic for the default traffic within its own policy map.
Figure 1 | Three Policy Maps Configured Without Fragments |
The figure below shows the same configuration configured with fragments and adds a fourth policy map with a class definition statement that classifies the fragments collectively. The default traffic classes are now classified as one service fragment group rather than three separate default traffic classes within the individual policy maps.
Figure 2 | Three Policy Maps Configured Using Fragments |
Understanding Fragments for Gigabit Etherchannel Bundles
Fragments can be configured for Gigabit Etherchannels when all of the member links of the Gigabit Etherchannel (GEC) bundle are on the same physical interface. Notably, if VLANs on the same physical interface are bundled, fragments can be used to define the collective treatment of all default traffic for the GEC bundle of VLAN subinterface member links.
When fragments are configured for Gigabit Etherchannel bundles, the policy maps that have a default traffic class configured using the fragment keyword are attached to the member subinterface links, and the policy maps that have a traffic class configured with the service-fragment keyword to collectively classify the fragments is attached to the physical interface.
Understanding the QoS Policies Aggregation MQC
The QoS: QoS: Policies Aggregation MQC Support for Multiple Queue Aggregation at Main Interface feature extends the previous support of aggregation of class-default traffic using the fragment and service-fragment configurations, to other user-defined traffic classes in a subinterface policy-map, such as DSCP-based traffic classes, that are aggregated at the main interface policy-map as shown in the figure below.
When no queueing is configured on a traffic class in the subinterface policy map, the account command can be used to track queueing drops that occur at the aggregate level for these classes, and can be displayed using the show policy-map interface command.
Figure 3 | Policy Map Overview for the MQC Support for Multiple Queue Aggregation at Main Interface Feature |
- Differences Between the Original Feature and the MQC Support for Multiple Queue Aggregation
- Changes in Queue Limit and WRED Thresholds
Differences Between the Original Feature and the MQC Support for Multiple Queue Aggregation
Although some of the configuration between the original QoS policies aggregation feature and enhancements in the MQC Support for Multiple Queue Aggregation at Main Interface feature appears similar, there are some important differences in the queueing behavior and the internal data handling.
For example, both configurations share and require the use of the fragment keyword for the class class-default command in the subscriber policy-map, as well as configuration of the service-fragmentkeyword for a user-defined class in the main interface policy-map to achieve common policy treatment for aggregate traffic. However, the use of this configuration results in different behavior between the original and enhanced QoS policies aggregation implementation:
- In the original implementation using the fragment and service-fragment architecture, all default class traffic and any traffic for classes without defined queueing features at the subinterface goes to the class-default queue and is aggregated into a common user-defined queue and policy defined at the main policy-map. Subinterface traffic aggregation (for example, from multiple subscribers on the same physical interface) ultimately occurs only for a single class, which is the default class.
- In the enhanced implementation of the MQC Support for Multiple Queue Aggregation at Main Interface feature also using the fragment and service-fragment architecture, all default class traffic also goes to the class-default queue and is aggregated into a common user-defined queue and policy defined at the main policy-map. However, other classes, such as DSCP-based subscriber traffic classes, are also supported for an aggregate policy. These traffic classes do not support any queues or queueing features other than account at the subscriber policy-map. The use of the fragment and service-fragment architecture enables these other subscriber traffic classes (from multiple subscribers on the same physical interface) to achieve common policy treatment for aggregate traffic that is defined for those same classes at the main policy-map.
The following sections summarize the key behavioral differences between the original QoS: Policies Aggregation feature and the QoS: QoS: Policies Aggregation MQC Support for Multiple Queue Aggregation at Main Interface feature.
Qos: Policies Aggregation Feature Prior to Cisco IOS XE Release 2.6
- All subinterface traffic classes have queues. However, when a traffic class in the subinterface policy-map is not configured with any queueing feature (commands such as priority, shape, bandwidth, queue-limit, fair-queue, random-detect, and so on, are not configured), the traffic is assigned to the class-default queue.
- Default class traffic from multiple subinterfaces can be aggregated into a common policy-map at the main interface when you use the fragment keyword at the subinterface class class-default configuration, and service-fragment configuration at the main interface class.
- No classification occurs or is supported at the main interface policy-map for any subinterface traffic classes that do not use the fragment and service-fragment configuration.
- Queueing occurs at the subinterface for other traffic classes defined with queueing features in the subinterface policy-map.
QoS: Policies Aggregation - MQC Support for Multiple Queue Aggregation at Main Interface Feature Beginning in Cisco IOS XE Release 2.6
- Subinterface traffic classes without configured queueing features do not have queues at the subscriber level.
- Default class traffic from multiple subinterfaces can be aggregated into a common policy-map at the main interface when you use the fragment keyword at the subinterface class class-default configuration, and service-fragment configuration at the main interface class. This configuration additionally enables support for other subinterface traffic classes (such as DSCP-based classes) to be aggregated into a common policy-map at the main interface.
- Other class traffic from multiple subinterfaces can be aggregated into a common policy-map at the main interface, according to the following configuration requirements:
- You enable this behavior by using the fragment keyword at the subinterface class class-default configuration, and service-fragment configuration at the main interface class (this also enables aggregation of the default class).
- You do not configure any queueing features at the subinterface policy-map for the other traffic classes.
- Queueing occurs at the main interface policy-map for other subinterface traffic classes as an aggregate.
- Optional tracking of statistics is supported using the account command for other traffic classes in the subinterface policy-map.
Changes in Queue Limit and WRED Thresholds
In Cisco IOS XE Release 2.6 the Cisco ASR 1000 Series Aggregation Services Routers support the addition of bytes as a unit of configuration for both queue limits and WRED thresholds. Therefore, as of this release, packet-based and byte-based limits are configurable, with some restrictions.
How to Configure QoS Policies Aggregation
- Configuring QoS Policies Aggregation for an Interface
- Configuring QoS Policies Aggregation on Gigabit Etherchannels
Configuring QoS Policies Aggregation for an Interface
Configuring a Fragment Traffic Class in a Policy Map
This procedure only shows how to configure the default traffic class as a fragment within a policy map. It does not include steps on configuring other classes within the policy map, or other policy maps on the router.
Like any policy map, the configuration is not managing network traffic until it has been attached to an interface. This procedure does not cover the process of attaching a policy map to an interface.
Note the following points about attaching and removing a policy map:
- To configure QoS: Policies Aggregation, you must attach the policy map that contains the service-fragment keyword to the main interface first, and then you must attach the policy map that contains the fragment keyword to the subinterface.
- To disable QoS: Policies Aggregation, you must remove the policy map that contains the fragment keyword from the subinterface first, and then you must remove the policy map that contains the service-fragment keyword from the main interface.
Note |
Only the default class statement in a policy map can be configured as a fragment. Fragments only work when multiple policy maps are attached to the same physical interface. This process cannot be used to classify default traffic classes as fragments on policy maps on different physical interfaces. Only queueing features are allowed in classes where the fragment keyword is entered, and at least one queueing feature must be entered in classes where the fragment keyword is used. A policy map with a class using the fragment keyword can only be applied to traffic leaving the interface (policy maps attached to interfaces using the service-policy output command). The fragment keyword cannot be entered in a child policy map. > |
DETAILED STEPS
Example
In the following example, a fragment named BestEffort is created in policy map subscriber1 and policy map subscriber 2. In this example, queueing features for other traffic classes are supported at the subinterface policy map.
policy-map subscriber1 class voice set cos 5 priority level 1 class video set cos 4 priority level 2 class class-default fragment BestEffort shape average 200000000 bandwidth remaining ratio 10 policy-map subscriber 2 class voice set cos 5 priority level 1 class video set cos 4 priority level 2 class class-default fragment BestEffort shape average 200000000 bandwidth remaining ratio 10
The following example also shows how to configure a fragment named BestEffort for the default class in a policy map on a subinterface using the QoS: QoS: Policies Aggregation MQC Support for Multiple Queue Aggregation at Main Interface implementation. In this example, notice that queueing features are not supported for the other classes in the policy map:
policy-map subscriber1 class voice set cos 5 account class video set cos 4 account class AF1 account class class-default fragment BestEffort shape average 200000000 bandwidth remaining ratio 10
What to Do Next
After configuring default class statements as fragments in multiple subinterface policy maps, a separate policy map with a class statement using the service-fragment keyword must be configured to apply QoS to the class statements configured as fragments.
This process is documented in the Configuring a Service Fragment Traffic Class.
Configuring a Service Fragment Traffic Class
This procedure assumes that fragment default traffic classes were already created. The procedure for creating fragment default traffic classes is documented in the Configuring a Fragment Traffic Class in a Policy Map.
Like any policy map, the configuration is not managing network traffic until it has been attached to an interface. This procedure does not cover the process of attaching a policy map to an interface.
Note |
A service fragment can be used to collectively classify fragments only from the same physical interface. Fragments from different interfaces cannot be classified using the same service fragment. Only queueing features are allowed in classes where the service-fragment keyword is entered, and at least one queueing feature must be entered in classes when the service-fragment keyword is used. A policy map with a class using the service-fragment keyword can only be applied to traffic leaving the interface (policy maps attached to interfaces using the service-policy output command). A class configured using the service-fragment keyword cannot be removed when it is being used to collectively apply QoS to fragments that are still configured on the interface. If you wish to remove a class configured using the service-fragment keyword, remove the fragment traffic classes before removing the service fragment. The service-fragment keyword cannot be entered in a child policy map. > |
DETAILED STEPS
Examples
In the following example, a policy map is created to apply QoS to all fragments named BestEffort.
policy-map main-interface class data service-fragment BestEffort shape average 400000000
In the following example, two fragments are created and then classified collectively using a service fragment.
policy-map subscriber1 class voice set cos 5 priority level 1 class video set cos 4 priority level 2 class class-default fragment BestEffort shape average 200000000 bandwidth remaining ratio 10 policy-map subscriber 2 class voice set cos 5 priority level 1 class video set cos 4 priority level 2 class class-default fragment BestEffort shape average 200000000 bandwidth remaining ratio 10
The following example shows the creation of two fragments called BestEffort in the subinterface policy maps, followed by a sample configuration for the service-fragment called BestEffort to aggregate the queues at the main interface policy map:
policy-map subscriber1 class voice set cos 5 account class video set cos 4 account class AF1 account class class-default fragment BestEffort shape average 200000000 bandwidth remaining ratio 10 policy-map subscriber2 class voice set cos 5 account class video set cos 4 account class AF1 account class class-default fragment BestEffort shape average 200000000 bandwidth remaining ratio 10 policy-map main-interface class voice priority level 1 class video priority level 2 class AF1 bandwidth remaining ratio 90 class data service-fragment BestEffort shape average 400000000 bandwidth remaining ratio 1
Troubleshooting Tips
Ensure that all class statements that are supposed to be part of the same service fragment share the same fragment-class-name.
What to Do Next
The policy map must be attached to an interface.
Configuring QoS Policies Aggregation on Gigabit Etherchannels
To properly configure QoS: Policies Aggregation on a Gigabit Etherchannel bundle, the following actions must be completed:
- Service fragment traffic classes must be configured and attached to the main physical interfaces.
- Fragment traffic classes must be configured and attached to the member link subinterfaces.
- Configuring Service Fragments on Physical Interface Supporting a Gigabit Etherchannel Bundle
- Configuring Fragments on Gigabit Etherchannel Member Link Subinterfaces
Configuring Service Fragments on Physical Interface Supporting a Gigabit Etherchannel Bundle
This procedure assumes that a service fragment traffic class has already been created. A service fragment traffic class cannot be configured without configuring a fragment class. The procedure for creating a fragment class is documented in the Configuring a Fragment Traffic Class in a Policy Map. The procedure for creating a service fragment traffic classes is documented in the Configuring a Service Fragment Traffic Class.
These instructions do not provide any details about the options that can be configured for Gigabit Etherchannel member link subinterfaces. These instructions only document the procedure for attaching a policy map that already has a fragment traffic class to a member link subinterface.
Note |
This process works only if all of the links of the GEC bundle are on the same physical interface. > |
DETAILED STEPS
Examples
Note |
This example shows a sample configuration that is supported for the original QoS: Policies Aggregation feature in releases prior to Cisco IOS XE Release 2.6. By following the newer policy-map configuration guidelines for the updates in Cisco IOS XE Release 2.6, it can be adapted to the QoS: QoS: Policies Aggregation MQC Support for Multiple Queue Aggregation at Main Interface feature. |
In the following example, policy map subscriber is configured with a fragment class named BE. The fragment is then configured as part of a policy map named aggregate-member-link. Policy map subscriber is then attached to the bundle subinterfaces while policy map aggregate-member-link is attached to the physical interface.
port-channel load-balancing vlan-manual class-map match any data ! class-map match-all BestEffort ! class-map match-all video ! class-map match-all voice ! policy-map subscriber class voice priority level 1 class video priority level 2 class class-default fragment BE shape average 100000000 bandwidth remaining ratios 80 policy-map aggregate-member-link class BestEffort service-fragment BE shape average 100000000 ! interface Port-channel1 ip address 10.0.0.0 255.255.0.0 ! interface Port-channel1.100 encapsulation dot1Q 100 ip address 10.0.0.1 255.255.255.0 service-policy output subscriber ! interface Port-channel1.200 encapsulation dot1Q 200 ip address 10.0.0.2 255.255.255.0 service-policy output subscriber ! interface Port-channel1.300 encapsulation dot1Q 300 ip address 10.0.0.4 255.255.255.0 service-policy output subscriber ! interface GigabitEthernet1/1/1 no ip address channel-group 1 mode on service-policy output aggregate-member-link ! interface GigabitEthernet1/1/2 no ip address channel-group 1 mode on service-policy output aggregate-member-link
Troubleshooting Tips
Ensure that the fragment-class-name is consistent across service-fragment and fragment class definitions.
What to Do Next
Attach the fragment service policy on the Gigabit Etherchannel member link subinterfaces.
Configuring Fragments on Gigabit Etherchannel Member Link Subinterfaces
This procedure assumes that a service fragment traffic class has already been created. A service fragment traffic class cannot be configured without configuring a fragment class. The procedure for creating a fragment class is documented in the Configuring a Fragment Traffic Class in a Policy Map. The procedure for creating a service fragment traffic classes is documented in the Configuring a Service Fragment Traffic Class.
These instructions do not provide any details about the options that can be configured for Gigabit Etherchannel member link subinterfaces. These instructions only document the procedure for attaching a policy map that already has a fragment traffic class to a member link subinterface.
Note |
Fragments cannot be used for traffic on two or more physical interfaces. The GEC must all be on the same physical interface for this configuration to work properly. > |
DETAILED STEPS
Example
Note |
This example shows a sample configuration that is supported for the original QoS: Policies Aggregation feature in releases prior to Cisco IOS XE Release 2.6. By following the newer policy-map configuration guidelines for the updates in Cisco IOS XE Release 2.6, it can be adapted to the QoS: QoS: Policies Aggregation MQC Support for Multiple Queue Aggregation at Main Interface feature. |
In the following example, the service policy named subscriber has a fragment default traffic class and is attached to the member link subinterface of a Gigabit Etherchannel bundle.
Note |
This example only shows how to attach a fragment default traffic class to the member link subinterface of a Gigabit Etherchannel bundle. This configuration is incomplete and would not classify default traffic appropriately until the physical interface was configured to support a service fragment traffic class. |
policy-map subscriber class voice priority level 1 class video priority level 2 class class-default fragment BE shape average 100000000 bandwidth remaining ratios 80 policy-map aggregate-member-link class BestEffort service-fragment BE shape average 100000000 ! interface Port-channel1 ip address 172.16.2.3 255.255.0.0 ! interface Port-channel1.100 encapsulation dot1Q 100 ip address 192.168.2.100 255.255.255.0 service-policy output subscriber !
Troubleshooting Tips
This configuration will not work until a service fragment default traffic class is created to classify the default traffic classes marked as fragments. This service fragment traffic class must be configured for this configuration to have any affect on network traffic.
What to Do Next
This is the final configuration step for configuring the QoS: Policies Aggregation feature on a Gigabit Etherchannel (GEC) bundle.
How to Configure QoS Policies Aggregation MQC
Some backward-compatibility exists between support of policies aggregation feature configuration in Cisco IOS XE Release 2.6 and prior Cisco IOS XE software releases. However, we recommend that you follow these upgrade guidelines for any physical interface where you want to move to the QoS: QoS: Policies Aggregation MQC Support for Multiple Queue Aggregation at Main Interface feature configuration.
For best results, you should upgrade any service policies configuration that you implemented prior to Cisco IOS XE Release 2.6, to the latest supported configuration.
The original and enhanced QoS: Policies Aggregation feature configuration can only reside on the same Cisco ASR 1000 Series Aggregation Services Router if the mixed configuration does not reside on the same physical interface. In other words, you can support the original configuration for one physical interface, and the enhanced configuration on a different physical interface.
The QoS: QoS: Policies Aggregation MQC Support for Multiple Queue Aggregation at Main Interface feature requires the same configuration of a fragment traffic class as the original feature, using the class class-default fragment command to enable and then define all subinterface policies aggregation, both for the default traffic class and the other traffic classes.
In the QoS: QoS: Policies Aggregation MQC Support for Multiple Queue Aggregation at Main Interface feature, the queueing features for the aggregate class queues (with traffic from the corresponding classes identified at the subinterfaces), are configured at the main interface policy-map.
- Upgrading Your Service Policies fo QoS Policies Aggregation - MQC
- Configuring QoS Policies Aggregation MQC Traffic Classes
- Configuring QoS Policies Aggregation MQC Support
- Verifying the Traffic Policy Class Policy Information and Drop Statistics
Upgrading Your Service Policies fo QoS Policies Aggregation - MQC
Prerequisites
Upgrading your service policies to support the QoS: QoS: Policies Aggregation MQC Support for Multiple Queue Aggregation at Main Interface feature assumes the following network conditions:
- The corresponding class-map statements appropriate for your network traffic are already configured.
- QoS service policies aggregation has been previously configured and applied for the main interface policy-map for a given physical interface and its corresponding subinterfaces, or subscriber interfaces, prior to Cisco IOS XE Release 2.6 for the default traffic class.
- A port on the same physical interface where you have previously configured the service policies aggregation feature prior to Cisco IOS XE Release 2.6 needs to support the configuration for the QoS: QoS: Policies Aggregation MQC Support for Multiple Queue Aggregation at Main Interface.
Upgrade Tasks
DETAILED STEPS
Step 1 | Configure the service policies for the QoS: QoS: Policies Aggregation MQC Support for Multiple Queue Aggregation at Main Interface feature. See the tasks described in the Configuring QoS Policies Aggregation MQC Traffic Classes. |
Step 2 | Remove any service policies configured prior to Cisco IOS XE Release 2.6 for any prior configured policies aggregation features using the no service-policy and no policy-map commands as follows:
|
Step 3 | Apply the new service policies for the QoS: QoS: Policies Aggregation MQC Support for Multiple Queue Aggregation at Main Interface feature at the appropriate interfaces using the service-policy output command as follows:
|
Configuring QoS Policies Aggregation MQC Traffic Classes
- Configuring Traffic Classes on the Subscriber Interface
- Configuring the Fragment Traffic Class on a Subinterface
- Configuring Traffic Classes at the Main Interface
- Configuring the Service Fragment Traffic Class at the Main Interface
Configuring Traffic Classes on the Subscriber Interface
DETAILED STEPS
Example
The following example configures the EF traffic class for policies aggregation at the subscriber subinterface with collection of drop statistics:
policy-map subscriber1 class EF account
What to Do Next
Follow this procedure for all traffic classes that you want to aggregate. Then, follow the instructions in the Configuring the Fragment Traffic Class on a Subinterface.
Configuring the Fragment Traffic Class on a Subinterface
What to Do Next
If you are upgrading your subinterface policy-map configuration from an earlier implementation of the QoS: Policies Aggregation feature, then remove the current service-policy from the subinterface using the no service-policy command.
Apply the new policy-map to outbound traffic on the subinterface using the service-policy output command.
Configuring Traffic Classes at the Main Interface
DETAILED STEPS
Command or Action | Purpose | |
---|---|---|
|
Example: Router> enable |
Enables privileged EXEC mode.
|
|
Example: Router# configure terminal |
Enters global configuration mode. |
|
Example: Router(config)# policy-map main-interface |
Specifies the name of the traffic policy to configure and enters policy map configuration mode. |
|
Example: Router(config-pmap)# class EF |
Specifies the name of the traffic class to be aggregated at the main interface policy-map, and enters policy-map class configuration mode. |
|
Example:
Example: Router(config-pmap-c)# priority level 1 |
Enters a QoS configuration command. The queueing features that are currently supported are bandwidth, priority, shape, and random-detect exponential-weighting-constant. Multiple QoS queueing commands can be entered. |
Example
The following example configures three traffic classes at the main interface policy-map, along with the aggregate service-fragment data class:
policy-map main-interface class voice priority level 1 class video priority level 2 class AF1 bandwidth remaining ratio 90 class data service-fragment BestEffort shape average 400000000 bandwidth remaining ratio 1
What to Do Next
Follow this procedure to define queueing features for all traffic classes that you want to aggregate. Then, follow the instructions in the Configuring the Service Fragment Traffic Class at the Main Interface.
Configuring the Service Fragment Traffic Class at the Main Interface
What to Do Next
If you are upgrading your main interface policy-map configuration from an earlier implementation of the QoS: Policies Aggregation feature, then remove the current service policy from the main interface using the no service-policy command.
Apply the new policy-map to outbound traffic on the main interface using the service-policy output command.
Configuring QoS Policies Aggregation MQC Support
The QoS: QoS: Policies Aggregation MQC Support for Multiple Queue Aggregation at Main Interface feature also supports configuration of the enhanced service policies on Gigabit Etherchannels according to the subscriber and main interface configuration guidelines described for this enhancement.
For more information, see the following sections:
Verifying the Traffic Policy Class Policy Information and Drop Statistics
To display information about policy-map configuration and subscriber drop statistics enabled using the account command, use the show policy-map interface command:
Router# show policy-map interface port-channel 1.1
Port-channel1.1
Service-policy input: input_policy
Class-map: class-default (match-any)
0 packets, 0 bytes
5 minute offered rate 0000 bps, drop rate 0000 bps
Match: any
QoS Set
dscp default
No packet marking statistics available
Service-policy output: Port-channel_1_subscriber
Class-map: EF (match-any)
105233 packets, 6734912 bytes
5 minute offered rate 134000 bps, drop rate 0000 bps
Match: dscp ef (46)
Match: access-group name VLAN_REMARK_EF
Match: qos-group 3
Account QoS statistics
Queueing
Packets dropped 0 packets/0 bytes
QoS Set
cos 5
No packet marking statistics available
dscp ef
No packet marking statistics available
Class-map: AF4 (match-all)
105234 packets, 6734976 bytes
5 minute offered rate 134000 bps, drop rate 0000 bps
Match: dscp cs4 (32)
Account QoS statistics
Queueing
Packets dropped 0 packets/0 bytes
QoS Set
cos 4
No packet marking statistics available
Class-map: AF1 (match-any)
315690 packets, 20204160 bytes
5 minute offered rate 402000 bps, drop rate 0000 bps
Match: dscp cs1 (8)
Match: dscp af11 (10)
Match: dscp af12 (12)
Account QoS statistics
Queueing
Packets dropped 0 packets/0 bytes
QoS Set
cos 1
No packet marking statistics available
Class-map: class-default (match-any) fragment Port-channel_BE
315677 packets, 20203328 bytes
5 minute offered rate 402000 bps, drop rate 0000 bps
Match: any
Queueing
queue limit 31250 bytes
(queue depth/total drops/no-buffer drops) 0/0/0
(pkts output/bytes output) 315679/20203482
bandwidth remaining ratio 1
Configuration Examples for QoS Policies Aggregation
- Example QoS Policies Aggregation
- Example Gigabit Etherchannel QoS Policies Aggregation
- Example QoS Policies Aggregation MQC Support at Main Interface
Example QoS Policies Aggregation
Note |
This example shows a sample configuration that is supported in the original QoS: Policies Aggregation feature prior to Cisco IOS XE Release 2.6. |
In the following example, QoS: Policies Aggregation is used to define a fragment class of traffic to classify default traffic using the default traffic class named BestEffort. All default traffic from the policy maps named subscriber1 and subscriber2 is part of the fragment default traffic class named BestEffort. This default traffic is then shaped collectively by creating a class called data that uses the service-fragment keyword and the shape command.
Note the following about this example:
- The class-name for each fragment default traffic class is "BestEffort."
- The class-name of "BestEffort" is also used to define the class where the service-fragment keyword is entered. This class applies a shaping policy to all traffic forwarded using the fragment default traffic classes named "BestEffort."
policy-map subscriber1 class voice set cos 5 priority level 1 class video set cos 4 priority level 2 class class-default fragment BestEffort shape average 200000000 bandwidth remaining ratio 10 policy-map subscriber 2 class voice set cos 5 priority level 1 class video set cos 4 priority level 2 class class-default fragment BestEffort shape average 200000000 bandwidth remaining ratio 10 policy-map input_policy class class-default set dscp default policy-map main-interface class data service-fragment BestEffort shape average 400000000 interface portchannel1.1001 encapsulation dot1q 1001 service-policy output subscriber1 service-policy input input_policy interface portchannel1.1002 encapsulation dot1q 1002 service-policy output subscriber2 service-policy input input_policy interface gigabitethernet 0/1 description member-link1 port channel 1 service-policy output main-interface interface gigabitethernet 0/2 description member-link2 port channel 1
service-policy output main-interface
Example Gigabit Etherchannel QoS Policies Aggregation
Note |
This example shows a sample configuration that is supported in the original QoS: Policies Aggregation feature prior to Cisco IOS XE Release 2.6. |
In the following example, policy map subscriber is configured with a fragment class named BE. The fragment is then configured as part of a policy map named aggregate-member-link. Policy map subscriber is then attached to the bundle subinterfaces while policy map aggregate-member-link is attached to the physical interface.
port-channel load-balancing vlan-manual class-map match-all BestEffort ! class-map match-all video ! class-map match-all voice ! policy-map subscriber class voice priority level 1 class video priority level 2 class class-default fragment BE shape average 100000000 bandwidth remaining ratios 80 policy-map aggregate-member-link class BestEffort service-fragment BE shape average 100000000 ! interface Port-channel1 ip address 10.1.1.3 255.255.0.0 ! interface Port-channel1.100 encapsulation dot1Q 100 ip address 10.1.2.1 255.255.255.0 service-policy output subscriber ! interface Port-channel1.200 encapsulation dot1Q 200 ip address 10.1.2.2 255.255.255.0 service-policy output subscriber ! interface Port-channel1.300 encapsulation dot1Q 300 ip address 10.1.2.3 255.255.255.0 service-policy output subscriber ! interface GigabitEthernet1/1/1 no ip address channel-group 1 mode on service-policy output aggregate-member-link ! interface GigabitEthernet1/1/2 no ip address channel-group 1 mode on service-policy output aggregate-member-link
Example QoS Policies Aggregation MQC Support at Main Interface
Note |
This example shows a sample configuration that is supported beginning in Cisco IOS XE Release 2.6. |
At the main interface policy map called Port-channel_1_main_policy, the queueing features for the DSCP-based subscriber traffic classes are configured. You can also see the use of byte-based queue limits and random-detect thresholds implemented at the main interface queues.
The service-fragment called Port-channel_BE is also configured to aggregate the traffic from the subscriber class-default fragment class.
policy-map Port-channel_1_main_policy class EF priority level 1 queue-limit 547500 bytes class AF4 priority level 2 queue-limit 4037500 bytes class AF1 bandwidth remaining ratio 90 queue-limit 750000 bytes random-detect dscp-based random-detect dscp 8 750000 bytes 750000 bytes random-detect dscp 10 750000 bytes 750000 bytes random-detect dscp 12 600000 bytes 675000 bytes class data service-fragment Port-channel_BE shape average 250000000 bandwidth remaining ratio 1 !
In this example, the policy map Port-channel_1_subscriber is configured with a fragment class named Port-channel_BE. (For simplicity, only a single subinterface policy is shown.) This enable queueing and policies aggregation for the subscriber traffic classes at the main interface policy map.
The Port-channel_1_subscriber policy map identifies the DSCP-based traffic classes of EF, AF4, and AF1 and enables collection of drop statistics for those classes.
policy-map Port-channel_1_subscriber class EF account set cos 5 set dscp ef class AF4 account set cos 4 class AF1 account set cos 1 class class-default fragment Port-channel_BE bandwidth remaining ratio 1 queue-limit 31250 bytes ! port-channel load-balancing vlan-manual ! interface Port-channel1 no ip address no negotiation auto !
The service policies are applied first to the physical interface, and then to the subinterfaces as shown:
interface GigabitEthernet1/2/0 no ip address negotiation auto no cdp enable service-policy output Port-channel_1_main_policy channel-group 1 ! interface GigabitEthernet2/2/0 no ip address negotiation auto service-policy output Port-channel_1_main_policy channel-group 1 ! interface Port-channel1.1 encapsulation dot1Q 2 primary GigabitEthernet1/2/0 secondary GigabitEthernet2/2/0 ip address 10.0.0.2 255.255.255.0 service-policy output Port-channel_1_subscriber
Additional References
Related Documents
Related Topic |
Document Title |
---|---|
Cisco IOS commands |
|
QoS commands: complete command syntax, command modes, command history, defaults, usage guidelines, and examples |
Cisco IOS Quality of Service Solutions Command Reference |
Modular Quality of Service Command-Line Interface |
"Applying QoS Features Using the MQC" module |
Distribution of Remaining Bandwidth Using Ratio |
"Distribution of Remaining Bandwidth Using Ratio" module |
Class-Based Shaping |
"Regulating Packet Flow-- Using Class-Based Traffic Shaping" module |
Standards
Standard |
Title |
---|---|
No new or modified standards are supported, and support for existing standards has not been modified by this feature. |
-- |
MIBs
MIB |
MIBs Link |
---|---|
CISCO-CLASS-BASED-QOS-MIB |
To locate and download MIBs for selected platforms, Cisco IOS XE software releases, and feature sets, use Cisco MIB Locator found at the following URL: |
RFCs
RFC |
Title |
---|---|
No new or modified RFCs are supported, and support for existing RFCs has not been modified. |
-- |
Technical Assistance
Description |
Link |
---|---|
The Cisco Support and Documentation website provides online resources to download documentation, software, and tools. Use these resources to install and configure the software and to troubleshoot and resolve technical issues with Cisco products and technologies. Access to most tools on the Cisco Support and Documentation website requires a Cisco.com user ID and password. |
Feature Information for QoS Policies Aggregation
The following table provides release information about the feature or features described in this module. This table lists only the software release that introduced support for a given feature in a given software release train. Unless noted otherwise, subsequent releases of that software release train also support that feature.
Use Cisco Feature Navigator to find information about platform support and Cisco software image support. To access Cisco Feature Navigator, go to www.cisco.com/go/cfn. An account on Cisco.com is not required.
Table 1 | Feature Information for QoS: Policies Aggregation |
Feature Name |
Releases |
Feature Information |
---|---|---|
QoS: Policies Aggregation |
Cisco IOS XE Release 2.1 |
This feature was introduced on Cisco ASR 1000 Series Routers. The following command was modified: class (policy-map). |
QoS: QoS: Policies Aggregation MQC Support for Multiple Queue Aggregation at Main Interface |
Cisco IOS XE Release 2.6 |
This feature was enhanced to support queueing aggregation at the primary interface for other traffic classes, including DSCP-based classes such as EF, AF1, and AF4 traffic classes. With this enhancement, other traffic classes from different subinterfaces share a common queue for that traffic class. Other enhancements include the ability to configure and show per-subscriber drop statistics on the aggregate queues and byte-based queue limits and WRED thresholds. In Cisco IOS XE Release 2.6, support for the CISCO-CLASS-BASED-QOS-MIB was added. The following commands are new or modified: account, show policy-map interface. |
Cisco and the Cisco logo are trademarks or registered trademarks of Cisco and/or its affiliates in the U.S. and other countries. To view a list of Cisco trademarks, go to this URL: www.cisco.com/go/trademarks. Third-party trademarks mentioned are the property of their respective owners. The use of the word partner does not imply a partnership relationship between Cisco and any other company. (1110R)
Any Internet Protocol (IP) addresses and phone numbers used in this document are not intended to be actual addresses and phone numbers. Any examples, command display output, network topology diagrams, and other figures included in the document are shown for illustrative purposes only. Any use of actual IP addresses or phone numbers in illustrative content is unintentional and coincidental.