Configuring QoS on the GGSN
This chapter describes how to configure Quality of Service (QoS) functions to differentiate traffic flow through the GGSN.
For a complete description of the GGSN commands in this chapter, refer to the Cisco IOS Mobile Wireless Command Reference. To locate documentation of other commands that appear in this chapter, use the command reference master index or search online.
This chapter includes the following sections:
•Overview of QoS Support on the GGSN
•Configuring Canonical QoS on the GGSN
•Configuring Delay QoS on the GGSN
•Configuring the GGSN Default QoS as Requested QoS
•Monitoring and Maintaining QoS on the GGSN
•Configuration Examples
Overview of QoS Support on the GGSN
GPRS QoS profiles are defined in the Global System for Mobile Communications (GSM) standards 02.60, 03.60, and 04.08. Each GPRS PDP context request contains a QoS profile. The QoS profile is considered a single parameter that defines the following data transfer class attributes according to the GSM standard:
•Precedence class
•Delay class
•Reliability class
•Peak throughput class
•Mean throughput class
The implementation of QoS support in the GPRS PLMN varies by the GPRS service provider and the available resources in the network. The GSM standards define the GPRS QoS classes that can be requested by an MS. However, the resulting QoS is negotiated and variable within the GPRS network backbone according to the implementations of the service provider.
This chapter describes the QoS support that the GGSN provides for the above GPRS QoS classes. As of GGSN Release 3.0, the GGSN adds a new method of QoS support—delay QoS. The GGSN currently supports the following two methods of QoS for GPRS traffic, only one of which can be activated globally on the GGSN for all GPRS traffic processing:
•Canonical QoS—Maps GPRS QoS classes to canonical QoS classes.
•Delay QoS—Maps GPRS QoS classes to delay QoS classes.
Configuring Canonical QoS on the GGSN
This section describes how to configure the canonical QoS method on the GGSN. It includes the following topics:
•Overview of Canonical QoS
•Canonical QoS Configuration Task List
•Verifying the Canonical QoS Configuration
•Canonical QoS Configuration Example
Overview of Canonical QoS
As of GGSN Release 1.2 and later, the Cisco Systems GGSN supports the canonical QoS method. The canonical QoS method on the GGSN supports three levels of QoS classification: Best effort, normal, and premium.
When you enable canonical QoS, the GGSN examines the QoS profile in PDP context requests for three of the five GPRS QoS classes (delay, precedence, and mean throughput). Based on combinations of values for those GPRS QoS class attributes, the GGSN maps the resulting QoS class to best effort, normal, or premium classifications.
Table 3 shows how the GGSN maps the different combinations of GPRS QoS class attributes within a PDP context request to a particular canonical QoS class, when canonical QoS is enabled on the GGSN. For example, if the QoS profile of a PDP context request specifies the best effort delay class, and any class of precedence and mean throughput, then the GGSN classifies that PDP context as the best effort canonical class.
Table 3 GPRS QoS Class Attribute Combinations Mapped to GGSN Canonical QoS Classes
|
|
|
|
Best effort |
Any |
Any |
Best effort |
1, 2, or 3 |
Low |
Any |
Best effort |
1, 2, or 3 |
Any |
Best effort |
Best effort |
1, 2, or 3 |
Normal |
Specified |
Normal |
1, 2, or 3 |
High |
Specified |
Premium |
Once you have enabled the canonical QoS method on the GGSN, you can map the canonical QoS classes to IP Type of Service (ToS) categories. IP ToS mappings allow the GGSN to support differentiated services according to RFC 2457. For more information, see the "Mapping Canonical QoS Classes to IP ToS Precedence" section. For more information about configuring the GGSN for differentiated services support, see the Cisco IOS Quality of Service Solutions Configuration Guide and Command Reference publications.
For the canonical QoS method, the GGSN sets aside a configurable amount of resource to be used for QoS processing. The GGSN allocates a portion of this total available resource for canonical QoS upon PDP context activation, based upon the QoS class to which the PDP context has been assigned. Typically, the GGSN uses more of its resources in support of the higher canonical QoS classes. As of GGSN Release 3.0, the total default amount of resource set aside by the GGSN for canonical QoS support is 3,145,728,000 bits per second. You can modify this value using the gprs canonical-qos gsn-resource-factor command. For more information, see the "Configuring Total GGSN Resources for Canonical QoS Support" section.
When a request for a user session comes in as a PDP context activation request, the GGSN determines whether the requested QoS for the session packets can be handled based on the amount of the gprs canonical-qos gsn-resource-factor that is available on the GGSN. Based on this determination, one of the following occurs:
•If the GGSN can provide the requested QoS, then the GGSN maintains that level of service.
•If the GGSN cannot provide the requested QoS, then the GGSN either lowers the QoS for the PDP context, or it rejects the PDP context request.
Canonical QoS Configuration Task List
To implement the canonical QoS method on the GGSN, you must enable the function. From there, you can modify the canonical QoS options to support your network environment.
To configure canonical QoS on the GGSN, perform the following tasks:
•Enabling Canonical QoS on the GGSN (Required)
•Mapping Canonical QoS Classes to IP ToS Precedence (Optional)
•Customizing the Canonical QoS Configuration (Optional)
Enabling Canonical QoS on the GGSN
Canonical QoS is not automatically enabled by the GGSN. To enable canonical QoS on the GGSN, use the following command in global configuration mode:
|
|
Router(config)# gprs qos map canonical-qos
|
Enables mapping of GPRS QoS categories to a canonical QoS method that includes best effort, normal, and premium canonical QoS classes. |
Mapping Canonical QoS Classes to IP ToS Precedence
Once you have enabled the canonical QoS method on the GGSN, you can map the canonical QoS classes to IP Type of Service (ToS) precedence. You can specify a mapping from the best effort, normal and premium canonical QoS categories to the ToS precedence bits (between 0 and 7, although 6 and 7 are not typically used). ToS precedence is reported in the IP header for packets transmitted over the Gn (GTP tunnel) and Gi interfaces.
All of the keyword arguments for the command are optional. However, if you specify a value for the normal argument, you must specify a value for the premium argument. And if you specify a value for the best-effort argument, then you must specify a value for both the premium and the normal arguments. The default Tos precedence values are 2 for premium, 1 for normal, and 0 for best effort.
The ToS precedence classes are defined as follows:
0 Routine
1 Priority
2 Immediate
3 Flash
4 Flash Override
5 Critical ECP
6 Internetwork Control
7 Network Control
Note The GTP signaling messages should always have the highest precedence in the GPRS network to help ensure the expedited delivery of those control messages. You can configure the ToS for GTP signaling messages using the gprs gtp map signalling tos command. The default value is 5.
To map canonical QoS classes to IP ToS precedence bits, use the following command in global configuration mode:
|
|
Router(config)# gprs canonical-qos map tos [premium tos-value [normal tos-value [best-effort tos-value]]]
|
(Optional) Specifies a QoS mapping from the canonical QoS classes to an IP ToS precedence value, where tos-value is an integer between 0 and 7 (values of 6 and 7 are not typically used). |
Customizing the Canonical QoS Configuration
This section describes some of the options that you can configure on the GGSN to further customize the default canonical QoS configuration.
Once you enable canonical QoS, the GGSN establishes default values for the allocation of GGSN resources to support canonical QoS processing. However, you most likely will want to modify the defaults based upon the GPRS traffic patterns and QoS profiles in use on your network.
This section includes the following topics:
•Configuring Total GGSN Resources for Canonical QoS Support
•Configuring GGSN Resources for the Best Effort Class
•Configuring the Deviation Factor for the Premium Class
Configuring Total GGSN Resources for Canonical QoS Support
For the canonical QoS method, the GGSN sets aside a configurable amount of resource that it uses for QoS processing. The GGSN allocates a portion of this total available resource for canonical QoS upon activating a PDP context, based upon the QoS class that the GGSN assigns to the PDP context. Typically, the GGSN uses more of its resources in support of the higher canonical QoS classes.
The GGSN allocates a portion of the total resource, and deducts that portion from the total available resource on the GGSN, according to the canonical QoS classes as follows:
•Best effort—The GGSN allocates the amount of resource specified by the gprs canonical-qos best-effort bandwidth-factor command for a best-effort PDP context. The default is 10 bps.
•Normal—The GGSN allocates the amount of resource according to the mean throughput value requested in the PDP context.
•Premium—The GGSN allocates the amount of resource according to a calculation of the minimum value of the requested peak throughput and mean throughput in the PDP context, along with a configurable deviation factor. You can configure the deviation factor using the gprs canonical-qos premium mean-throughput-deviation command.
Once the GGSN allocates resources for a PDP context, it does not make the resource available again until it deletes the PDP context or it receives an update request that requires a change to the allocated resource.
The total default amount of resource set aside by the GGSN for canonical QoS support is 3,145,728,000 bits per second. The default value for this command was chosen to support 10000 PDP contexts with a premium QoS class. If you require greater throughput for the GPRS data on your network, increase the resource factor value. However, be aware that if you select a value that is too high, you might exceed the actual processing capacity of the GGSN.
To configure the total GGSN resource for canonical QoS support, use the following command in global configuration mode:
|
|
Router(config)# gprs canonical-qos gsn-resource-factor resource-factor |
(Optional) Specifies the total amount of resource that the GGSN uses to provide QoS service levels to mobile users. The default is 3,145,728,000 bits per second. |
Configuring GGSN Resources for the Best Effort Class
You can also configure resource to be reserved for best effort QoS classes on the GGSN using the gprs canonical-qos best-effort bandwidth-factor command. This command specifies an average bandwidth that is expected to be used by best-effort QoS class mobile sessions. The default value is 10 bps. If you observe that users accessing the GGSN are using a higher average bandwidth, then you should increase the bandwidth value.
To modify the bandwidth factor for the best-effort canonical QoS class, use the following command in global configuration mode:
|
|
Router(config)# gprs canonical-qos best-effort bandwidth-factor bandwidth-factor
|
(Optional) Specifies the bandwidth factor to be applied to the canonical best-effort QoS class. The default value is 10 bps. |
Configuring the Deviation Factor for the Premium Class
The GGSN uses the minimum value of the requested peak throughput and mean throughput in the PDP context, along with a configurable deviation factor to determine how much resource to allocate for the premium QoS class.
You can configure a deviation factor (factor/1000) to adjust the result of the calculation that the GGSN uses to determine the amount of data throughput to allocate for premium QoS support.
The GGSN bases its calculation on the following formula, which includes the throughput deviation factor:
EB = Min[p, m + a (p - m)]
Where
•EB = the effective bandwidth
•p = peak throughput from the GPRS QoS profile in the PDP context request
•m = mean throughput from the GPRS QoS profile in the PDP context request
•a = the deviation factor, a, divided by 1000 (a/1000)
To configure the deviation factor that the GGSN uses for calculation of premium canonical QoS support, use the following command in global configuration mode:
|
|
Router(config)# gprs canonical-qos premium mean-throughput-deviation deviation-factor |
(Optional) Specifies a mean throughput deviation factor that the GGSN uses to calculate the allowable data throughput for the premium QoS class. The default is 100. |
Verifying the Canonical QoS Configuration
To verify your canonical QoS configuration, use the show running-config command and observe the canonical QoS parameters as shown in bold in the following example:
Router# show running-config
Building configuration...
Current configuration : 3521 bytes
no service single-slot-reload-enable
service timestamps debug uptime
service timestamps log uptime
no service password-encryption
no ip dhcp-client network-discovery
ip address 10.100.3.4 255.255.255.255
interface FastEthernet0/0
ip address 172.18.43.174 255.255.255.240
description Gi interface to gprt.cisco.com
ip address 10.8.8.6 255.255.255.0
description Gi interface to gprs.cisco.com
ip address 10.9.9.4 255.255.255.0
ip address 10.15.15.10 255.255.255.0
interface Virtual-Template1
ip address 10.40.40.3 255.255.255.0
gprs access-point-list gprs
gprs access-point-list gprs
access-mode non-transparent
access-point-name www.pdn1.com
aaa-group authentication foo
access-mode non-transparent
access-point-name www.pdn2.com
access-point-name www.pdn4.com
aaa-group accounting foo1
access-point-name www.pdn5.com
gprs maximum-pdp-context-allowed 90000
gprs qos map canonical-qos
gprs canonical-qos gsn-resource-factor 4294967295
gprs canonical-qos best-effort bandwidth-factor 10000
gprs canonical-qos premium mean-throughput-deviation 500
gprs canonical-qos map tos premium 3 normal 2 best-effort 1
gprs gtp path-echo-interval 30
gprs default aaa-group authentication foo2
gprs default aaa-group accounting foo3
gprs default charging-gateway 10.15.15.1
Configuring Delay QoS on the GGSN
This section describes how to configure the delay QoS method on the GGSN. It includes the following topics:
•Overview of Delay QoS
•Delay QoS Configuration Task List
•Verifying the Delay QoS Configuration
•Delay QoS Configuration Example
Overview of Delay QoS
As of GGSN Release 3.0 and later, the Cisco Systems GGSN supports the delay QoS method. The delay QoS method on the GGSN supports four levels of QoS classification: Class 1, class 2, class 3 and best effort.
When you enable delay QoS, the GGSN examines the QoS profile in PDP context requests for three of the five GPRS QoS classes (delay, precedence, and mean throughput). Based on combinations of values for those GPRS QoS class attributes, the GGSN maps the resulting delay QoS class to class 1, class 2, class 3, or best effort categories.
Table 4 shows how the GGSN maps the different combinations of GPRS QoS class attributes within a PDP context request to a particular delay QoS class, when delay QoS is enabled on the GGSN. For example, if the QoS profile of a PDP context request specifies the best effort delay class, and any class of precedence and mean throughput, then the GGSN classifies that PDP context as the best effort delay class.
Table 4 GPRS QoS Class Attribute Combinations Mapped to GGSN Delay QoS Classes
|
|
|
|
Undefined |
Any |
Any |
Best effort |
Best effort |
Any |
Any |
Best effort |
Class 1 |
Any |
Any |
Class 1 |
Class 2 |
Any |
Any |
Class 2 |
Class 3 |
Any |
Any |
Class 3 |
Delay QoS Configuration Task List
To implement the delay QoS method on the GGSN, you must enable the function. From there, you can modify the delay QoS options to support your network environment.
To configure delay QoS on the GGSN, perform the following tasks:
•Enabling Delay QoS on the GGSN (Required)
•Mapping Delay QoS Classes to IP ToS Precedence (Optional)
Enabling Delay QoS on the GGSN
Delay QoS is not automatically enabled by the GGSN. To enable delay QoS on the GGSN, use the following command in global configuration mode:
|
|
Router(config)# gprs qos map delay
|
Enables mapping of GPRS QoS categories to a delay QoS method that includes the class 1, class 2, class 3, and best effort classes. |
Mapping Delay QoS Classes to IP ToS Precedence
Once you have enabled the delay QoS method on the GGSN, you can map the delay QoS classes to IP Type of Service (ToS) precedence. You can specify a mapping from the class1, class2, class3, or class best effort delay QoS categories to the ToS precedence bits (between 0 and 7, although 6 and 7 are not typically used). ToS precedence is reported in the IP header for packets transmitted over the Gn (GTP tunnel) and Gi interfaces.
The class2, class3 and class-best-effort keyword arguments are optional. However, if you specify a value for the class3 argument, you must specify a value for the class2 argument. And, if you specify a value for the class-best-effort argument, then you must specify a value for both the class2 and the class3 arguments.
The ToS precedence classes are defined as follows:
0 Routine
1 Priority
2 Immediate
3 Flash
4 Flash Override
5 Critical ECP
6 Internetwork Control
7 Network Control
Note The GTP signaling messages should always have the highest precedence in the GPRS network to help ensure the expedited delivery of those control messages. You can configure the ToS for GTP signaling messages using the gprs gtp map signalling tos command. The default value is 5.
To map delay QoS classes to IP ToS precedence bits, use the following command in global configuration mode:
|
|
Router(config)# gprs delay-qos map tos class1 tos-value [class2 tos-value [class3 tos-value [class-best-effort tos-value]]]
|
(Optional) Specifies a QoS mapping from the delay QoS classes to an IP ToS precedence value, where tos-value is an integer between 0 and 5 (values of 6 and 7 are not typically used). |
Verifying the Delay QoS Configuration
To verify your delay QoS configuration, use the show running-config command and observe the delay QoS parameters as shown in bold in the following example:
Router# show running-config
Building configuration...
Current configuration : 3521 bytes
no service single-slot-reload-enable
service timestamps debug uptime
service timestamps log uptime
no service password-encryption
no ip dhcp-client network-discovery
ip address 10.100.3.4 255.255.255.255
interface FastEthernet0/0
ip address 172.18.43.174 255.255.255.240
description Gi interface to gprt.cisco.com
ip address 10.8.8.6 255.255.255.0
description Gi interface to gprs.cisco.com
ip address 10.9.9.4 255.255.255.0
ip address 10.15.15.10 255.255.255.0
interface Virtual-Template1
ip address 10.40.40.3 255.255.255.0
gprs access-point-list gprs
gprs access-point-list gprs
access-mode non-transparent
access-point-name www.pdn1.com
aaa-group authentication foo
access-mode non-transparent
access-point-name www.pdn2.com
access-point-name www.pdn4.com
aaa-group accounting foo1
access-point-name www.pdn5.com
gprs maximum-pdp-context-allowed 45000
gprs delay-qos map tos class1 4 class2 3 class3 2 class-best-effort 1
gprs gtp path-echo-interval 30
gprs default aaa-group authentication foo2
gprs default aaa-group accounting foo3
gprs default charging-gateway 10.15.15.1
Configuring the GGSN Default QoS as Requested QoS
If you are not using canonical or delay based QoS mapping on the GGSN, you can configure the GGSN to set its default QoS values in the response message exactly as requested in the create PDP context request message. By using this command, you can prevent the GGSN from lowering the requested QoS.
To configure the GGSN to set the requested QoS as the default QoS, use the following command beginning in global configuration mode:
|
|
Router(config)# gprs qos default-response requested
|
(Optional) Specifies that the GGSN sets its default QoS values in the response message exactly as requested in the create PDP context request message. |
Note When the gprs qos default-response requested command is not configured, and canonical QoS is not enabled, the GGSN sets its default QoS class to best effort.
Monitoring and Maintaining QoS on the GGSN
This section describes the commands used to display QoS configuration parameters and status on the GGSN.
It includes the following topics:
•Show Command Summary
•Displaying QoS Information for a PDP Context
•Displaying Canonical QoS Status on the GGSN
•Displaying PDP Contexts by Canonical QoS Precedence Class
•Displaying Delay QoS Status on the GGSN
•Displaying PDP Contexts by Delay QoS Class
Show Command Summary
This section provides a summary list of the show commands that you can use to monitor canonical or delay QoS on the GGSN. Not all commands provide information for both types of QoS methods on the GGSN.
The following privileged EXEC commands are used to monitor and maintain QoS on the GGSN:
|
|
Router# show gprs gtp pdp-context imsi hex-data
|
Displays PDP contexts by International Mobile Subscriber Identity (IMSI). |
Router# show gprs gtp pdp-context qos-delay {class1 |
class2 | class3 | classbesteffort}
|
Displays PDP contexts for a specified GPRS quality of service delay class type. (Applicable to delay QoS only.) |
Router# show gprs gtp pdp-context qos-precedence {low
| normal | high}
|
Displays PDP contexts for a specified GPRS quality of service precedence type. (Applicable to canonical QoS only.) |
Router# show gprs gtp pdp-context tid hex-data
|
Displays PDP contexts by tunnel ID. |
Router# show gprs gtp status
|
Displays information about the current status of GTP on the GGSN. (Includes either canonical or delay QoS information.) |
Displaying QoS Information for a PDP Context
To display QoS information for a particular PDP context, you can use the show gprs gtp pdp-context command using the tid or imsi keywords. The following example shows sample output for the show gprs gtp pdp-context tid command for a PDP context in the best effort canonical QoS class (canonical QoS class(neg)=01). The output fields displaying QoS information are shown in bold:
router#show gprs gtp pdp tid 1111111111111111
TID MS Addr Source SGSN Addr APN
1111111111111111 10.1.1.1 Radius 10.8.8.1 dns.com
current time :Mar 18 2002 11:24:36
user_name (IMSI):1111111111111111 MS address:10.1.1.1
MS International PSTN/ISDN Number (MSISDN):ABC
sgsn_addr_signal:10.8.8.1 ggsn_addr_signal:10.8.0.1
signal_sequence: 0 seq_tpdu_up: 0
upstream_signal_flow: 1 upstream_data_flow: 2
downstream_signal_flow:14 downstream_data_flow:12
pdp_create_time: Mar 18 2002 09:58:39
last_access_time: Mar 18 2002 09:58:39
mnrgflag: 0 tos mask map:20
gprs qos_req:091101 canonical Qos class(req.):01
gprs qos_neg:25131F canonical Qos class(neg.):01
rcv_byte_count: 0 rcv_pkt_count: 0
send_byte_count: 0 send_pkt_count: 0
cef_up_pkt: 0 cef_up_byte: 0
cef_down_pkt: 0 cef_down_byte: 0
** Network Init Information **
MNRG Flag: 0 PDU Discard Flag: 0
SGSN Addr: 172.16.44.1 NIP State: NIP_STATE_WAIT_PDP_ACTIVATION
Note The canonical QoS class and effective bandwidth output fields only apply when canonical QoS is in use on the GGSN.
The following sections describe how you can interpret some of the QoS information that is provided by the show gprs gtp pdp-context command:
•Determining the ToS Precedence
•Interpreting the Requested and Negotiated GPRS QoS
•Interpreting the Effective Bandwidth for a PDP Context (Canonical QoS only)
Determining the ToS Precedence
To determine the ToS precedence for a PDP context, you need to convert the hexadecimal value shown in the tos mask map output field of the show gprs gtp pdp-context command to binary format. From there, you can interpret the ToS precedence bits, which are the first 3 bits of the binary conversion.
In the following example, we use a tos mask map value of 20 to show this conversion:
Step 1 Convert the value of the tos mask map field (20) to binary, where 2=0010 and 0=0000. This results in the following binary format:
0010 0000
Step 2 Identify the first 3 bits of the binary representation, which is 001-0 in our example. (The remaining 0000 bits are ignored.)
Step 3 Convert the first 3 bits to a decimal number. In our example, 001=1. Therefore, the ToS precedence for this PDP context is 1.
Interpreting the Requested and Negotiated GPRS QoS
To determine the various GPRS QoS class attributes shown in the gprs qos_req and gprs qos_neg output fields of the show gprs gtp pdp-context command, you need to convert the values provided to binary format. From there, you can interpret the class attribute values according to the GSM specifications for QoS, which can be found in GSM standards 02.60, 03.60, and 04.08.
In the following example, we use a GPRS QoS value of 25131F to show this conversion:
Step 1 Convert the hexadecimal value of the gprs qos_req or gprs qos_neq field (25131F) to binary, where 2=0010, 5=0101, 1=0001, 3=0011, 1=0001, and F=1111. This results in the following binary format:
0010 0101 0001 0011 0001 1111
Step 2 Group the bits in the following manner:
First 2 bits |
Next 3 bits |
Next 3 bits |
Next 4 bits |
Next 1 bit |
Next 3 bits |
Next 3 bits |
Last 5 bits |
00 |
100 |
101 |
0001 |
0 |
011 |
000 |
1 1111 |
don't care |
delay |
reliability |
peak |
don't care |
precedence |
don't care |
mean throughput |
Step 3 Convert the bit groups to decimal numbers, and correlate the value to the QoS classes according to the GSM specifications. For example, for the delay class, the binary 100=4, which corresponds to delay class 4.
In this example, the corresponding QoS classes are delay class 4, reliability class 5, peak class 1, precedence class 3, and mean throughput is best effort:
First 2 bits |
Next 3 bits |
Next 3 bits |
Next 4 bits |
Next 1 bit |
Next 3 bits |
Next 3 bits |
Last 5 bits |
00 |
100 |
101 |
0001 |
0 |
011 |
000 |
1 1111 |
don't care |
delay |
reliability |
peak |
don't care |
precedence |
don't care |
mean throughput |
|
class 4 |
class 5 |
class 1 |
|
class 3 |
|
best effort |
Interpreting the Effective Bandwidth for a PDP Context
You can use the show gprs gtp pdp-context tid command to display an output field called effective bandwidth in bits per second. The effective bandwidth is determined according to the canonical QoS class (premium, normal, or best effort) for the PDP context. However, it is an estimate and does not represent the actual bandwidth in use by the PDP context.
You can calculate the potential number of supported PDP contexts for a class of QoS using the effective bandwidth value. To determine an estimate of the potential number of PDP contexts of a particular class that can be supported on the GGSN, you can divide the total bandwidth available on the GGSN by the effective bandwidth value for the QoS class.
The following example shows how to estimate the potential number of PDP contexts that the GGSN can support for a particular canonical QoS class at an expected effective bandwidth:
Step 1 Use the show gprs gtp pdp-context command with either the tid or imsi keywords and find the value of the effective bandwidth field. In our example, we will use 10000 bps.
Step 2 To estimate the number of best effort PDP contexts that the GGSN can support with an effective bandwidth of 10000 bps, divide the total amount of resource on the GGSN for canonical QoS by the effective bandwidth used.
In this example, we will use the default total resource value of 4294967295 and the following calculation:
where 4294957295 is the total resource. The result is an estimated 429496 best effort PDP contexts.
Note To verify the total amount of resource on the GGSN for canonical QoS, you can use the show gprs gtp status command.
Displaying Canonical QoS Status on the GGSN
You can use the show gprs gtp status command to display several different types of canonical QoS information, including GGSN resources in use, number of active PDP contexts by canonical QoS class, and mean throughput by canonical QoS class.
Note The output of the show gprs gtp status command varies depending on the type of QoS method in use on the GGSN.
The following example shows 2 active PDP contexts on the GGSN that are using the best effort canonical QoS class. The mean throughput for the 2 PDP contexts is 20000 bps (a cumulative value, which corresponds to an effective bandwidth of 10000 bps for each PDP context in this example):
Router# show gprs gtp status
gsn_used_bandwidth 20000 total gsn_resource 4294967295
activated_pdp 2 ntwk_init_pdp 0
activated_ppp_pdp 0 activated_ppp_regen_pdp 0
mean_throughput_premium 0.000
mean_throughput_normal 0.000 mean_throughput_besteffort 20000.000
qos_high_pdp 0 qos_normal_pdp 0
qos_low_pdp 2 qos premium mean-throughput-deviation 0.500
Interpreting the GGSN Resources Allocated for Canonical QoS Support
The show gprs gtp status command shows cumulative values for the currently active PDP contexts on the GGSN (the total gsn_resource and qos premium mean-throughput-deviation values are not cumulative).
For multiple PDP contexts, the used resource is a cumulative value across all active PDP contexts and can represent different QoS classes. In the example, the gsn_used_bandwidth value of 20000 bps represents the total bps in use for the 2 best effort PDP contexts.
To determine the amount of available GGSN resource remaining for canonical QoS support, you can subtract the current value of the gsn_used_bandwidth from the total gsn_resource. In this example, the calculation is:
which equals an estimated 4294947295 resource remaining for canonical QoS processing.
Displaying PDP Contexts by Canonical QoS Precedence Class
To display the current number of active PDP contexts by canonical QoS precedence class, perform the following steps:
Step 1 To verify the canonical QoS precedence class for which there are currently active PDP contexts, use the show gprs gtp status command. The following example shows that 1 PDP context is currently active for the high precedence (or premium canonical QoS) class on the GGSN:
Router# show gprs gtp status
gsn_used_bandwidth 800 total gsn_resource 1048576
activated_pdp 1 ntwk_init_pdp 0
activated_ppp_pdp 0 activated_ppp_regen_pdp 0
mean_throughput_premium 0.220
mean_throughput_normal 0.000 mean_throughput_besteffort 0.000
qos_high_pdp 1 qos_normal_pdp 0
qos_low_pdp 0 qos premium mean-throughput-deviation 0.100
Step 2 To display information about active PDP contexts in a particular precedence class, use the show gprs gtp pdp-context qos-precedence command. The following example shows information about the active PDP context in the high precedence (premium) class:
Router# show gprs gtp pdp-context qos-precedence high
TID MS Addr Source SGSN Addr APN
4444444444444444 10.2.0.4 Static 10.39.39.1 www.pdn2.com
Displaying Delay QoS Status on the GGSN
To display the current number of active PDP contexts by delay QoS class, use the show gprs gtp status command. The following example shows 1 active PDP context using delay class 1, 1 active PDP context using delay class 2, and 2 active PDP contexts using the delay best effort class. The total number of 4 PDP contexts is indicated in the activated_pdp output field:
Router# show gprs gtp status
activated_pdp 4 ntwk_init_pdp 0
activated_ppp_pdp 0 activated_ppp_regen_pdp 0
qos_delay1_pdp 1 qos_delay2_pdp 1
qos_delay3_pdp 0 qos_delaybesteffort_pdp 2
Displaying PDP Contexts by Delay QoS Class
To display the current number of active PDP contexts by delay QoS class, perform the following steps:
Step 1 To verify the delay QoS classes for which there are currently active PDP contexts, use the show gprs gtp status command. The following examples shows that there are active PDP contexts for each of the delay classes except class 3:
Router# show gprs gtp status
activated_pdp 4 ntwk_init_pdp 0
activated_ppp_pdp 0 activated_ppp_regen_pdp 0
qos_delay1_pdp 1 qos_delay2_pdp 1
qos_delay3_pdp 0 qos_delaybesteffort_pdp 2
Step 2 To display information about PDP contexts in a particular delay class, use the show gprs gtp pdp-context qos-delay command as shown in the following examples:
Example 1
The following example shows information about the active PDP contexts in the best effort delay QoS class:
Router# show gprs gtp pdp-context qos-delay classbesteffort
TID MS Addr Source SGSN Addr APN
1111111111111111 10.8.8.1 Static 10.39.39.1 gprt.cisco.com
2222222222222222 10.8.8.2 Static 10.39.39.1 gprt.cisco.com
Example 2
The following example shows information about the active PDP context in delay class 1:
Router# show gprs gtp pdp-context qos-delay class1
TID MS Addr Source SGSN Addr APN
3333333333333333 10.8.8.4 Static 10.39.39.1 gprt.cisco.com
Configuration Examples
This section includes the following examples:
•Canonical QoS Configuration Example
•Delay QoS Configuration Example
Canonical QoS Configuration Example
The following example shows part of a sample GGSN configuration for the canonical QoS method:
Router# show running-config
Building configuration...
Current configuration : 3521 bytes
no service single-slot-reload-enable
service timestamps debug uptime
service timestamps log uptime
no service password-encryption
no ip dhcp-client network-discovery
ip address 10.100.3.4 255.255.255.255
interface FastEthernet0/0
ip address 172.18.43.174 255.255.255.240
description Gi interface to gprt.cisco.com
ip address 10.8.8.6 255.255.255.0
description Gi interface to gprs.cisco.com
ip address 10.9.9.4 255.255.255.0
ip address 10.15.15.10 255.255.255.0
interface Virtual-Template1
ip address 10.40.40.3 255.255.255.0
gprs access-point-list gprs
gprs access-point-list gprs
access-mode non-transparent
access-point-name gprt.cisco.com
aaa-group authentication foo
access-mode non-transparent
access-point-name gprs.cisco.com
access-point-name gpru.cisco.com
aaa-group accounting foo1
access-point-name gprv.cisco.com
gprs maximum-pdp-context-allowed 90000
gprs qos map canonical-qos
! Configure total resource available
! for canonical QoS processing
gprs canonical-qos gsn-resource-factor 4294967295
! Configure bandwidth estimated for
! best effort canonical QoS class
gprs canonical-qos best-effort bandwidth-factor 10000
! Configure deviation factor for mean throughput
! calculation for premium QoS class
gprs canonical-qos premium mean-throughput-deviation 500
! Configure ToS precedence mapping to
gprs canonical-qos map tos premium 3 normal 2 best-effort 1
gprs gtp path-echo-interval 30
gprs default aaa-group authentication foo2
gprs default aaa-group accounting foo3
gprs default charging-gateway 10.15.15.1
Delay QoS Configuration Example
The following example shows part of a sample GGSN configuration for the delay QoS method:
Router# show running-config
Building configuration...
Current configuration : 3521 bytes
no service single-slot-reload-enable
service timestamps debug uptime
service timestamps log uptime
no service password-encryption
no ip dhcp-client network-discovery
ip address 10.100.3.4 255.255.255.255
interface FastEthernet0/0
ip address 172.18.43.174 255.255.255.240
description Gi interface to gprt.cisco.com
ip address 10.8.8.6 255.255.255.0
description Gi interface to gprs.cisco.com
ip address 10.9.9.4 255.255.255.0
ip address 10.15.15.10 255.255.255.0
interface Virtual-Template1
ip address 10.40.40.3 255.255.255.0
gprs access-point-list gprs
gprs access-point-list gprs
access-mode non-transparent
access-point-name gprt.cisco.com
aaa-group authentication foo
access-mode non-transparent
access-point-name gprs.cisco.com
access-point-name gpru.cisco.com
aaa-group accounting foo1
access-point-name gprv.cisco.com
gprs maximum-pdp-context-allowed 45000
! Configure ToS precedence mapping to
gprs delay-qos map tos class1 4 class2 3 class3 2 class-best-effort 1
gprs gtp path-echo-interval 30
gprs default aaa-group authentication foo2
gprs default aaa-group accounting foo3
gprs default charging-gateway 10.15.15.1