Contents

Configuring PBR·· 1

About PBR· 1

Packet forwarding process· 1

PBR types· 1

Policy· 1

PBR and Track· 2

Restrictions and guidelines: PBR configuration· 3

PBR tasks at a glance· 3

Configuring a policy· 3

Creating a node· 3

Setting match criteria for a node· 3

Configuring actions for a node· 4

Specifying a policy for PBR· 8

Specifying a policy for local PBR· 8

Specifying a policy for interface PBR· 8

Display and maintenance commands for PBR· 9

PBR configuration examples· 9

Example: Configuring packet type-based local PBR· 9

Example: Configuring packet type-based interface PBR· 11

Example: Configuring packet length-based interface PBR· 12

Configuring PBR

About PBR

Policy-based routing (PBR) uses user-defined policies to route packets. A policy can specify parameters for packets that match specific criteria such as ACLs or that have specific lengths. The parameters include the next hop, output interface, default next hop, and default output interface.

Packet forwarding process

When the device receives a packet, the device searches the PBR policy for a matching node to forward that packet.

·     If a matching node is found and its match mode is permit, the device performs the following operations:

a.     Uses the next hops or output interfaces specified on the node to forward the packet.

b.     Searches the routing table for a route (except the default route) to forward the packet if one of the following conditions exists:

-     No next hops or output interfaces are specified on the node.

-     Forwarding failed based on the next hops or output interfaces.

c.     Uses the default next hop or default output interface specified on the node to forward the packet if one of the following conditions exists:

-     No matching route was found in the routing table.

-     The routing table-based forwarding failed.

d.     Uses the default route to forward the packet if one of the following conditions exists:

-     No default next hops or default output interfaces are specified on the node.

-     The forwarding failed based on the default next hops or default output interfaces.

·     The device perfoms routing table lookup to forward the packet in either of the following conditions:

¡     No matching node is found.

¡     A matching node is found, but its match mode is deny.

PBR types

PBR includes the following types:

·     Local PBR—Guides the forwarding of locally generated packets, such as ICMP packets generated by using the ping command.

·     Interface PBR—Guides the forwarding of packets received on an interface.

Policy

A policy includes match criteria and actions to be taken on the matching packets. A policy can have one or multiple nodes as follows:

·     Each node is identified by a node number. A smaller node number has a higher priority.

·     A node contains if-match and apply clauses. An if-match clause specifies a match criterion, and an apply clause specifies an action.

·     A node has a match mode of permit or deny.

A policy compares packets with nodes in priority order. If a packet matches the criteria on a node, it is processed by the action on the node. If the packet does not match any criteria on the node, it goes to the next node for a match. If the packet does not match the criteria on any node, the device performs a routing table lookup.

Relationship between if-match clauses

On a node, you can specify multiple types of if-match clauses but only one if-match clause for each type.

To match a node, a packet must match all types of the if-match clauses for the node but only one if-match clause for each type.

Relationship between apply clauses

You can specify multiple apply clauses for a node, but some of them might not be executed. For more information about relationship between apply clauses, see "Configuring actions for a node."

Relationship between the match mode and clauses on the node

Does a packet match all the if-match clauses on the node?	Match mode
	Permit	Deny
Yes.	·     If the node contains apply clauses, PBR executes the apply clauses on the node. ¡     If PBR-based forwarding succeeds, PBR does not compare the packet with the next node. ¡     If PBR-based forwarding fails and the apply continue clause is not configured, PBR does not compare the packet with the next node. ¡     If PBR-based forwarding fails and the apply continue clause is configured, PBR compares the packet with the next node. ·     If the node does not contain apply clauses, the device performs a routing table lookup for the packet.	The device performs a routing table lookup for the packet.
No.	PBR compares the packet with the next node.	PBR compares the packet with the next node.

‌

NOTE: A node that has no if-match clauses matches any packet.

‌

PBR and Track

PBR can work with the Track feature to dynamically adapt the availability status of an apply clause to the link status of a tracked object. The tracked object can be a next hop, output interface, default next hop, or default output interface.

·     When the track entry associated with an object changes to Negative, the apply clause is invalid.

·     When the track entry changes to Positive or NotReady, the apply clause is valid.

For more information about Track-PBR collaboration, see Network Management and Monitoring Configuration Guide.

Restrictions and guidelines: PBR configuration

If the device performs forwarding in software, PBR does not process IP packets destined for the local device.

If the device performs forwarding in hardware and a packet destined for it matches a PBR policy, PBR will execute the apply clauses in the policy, including the clause for forwarding. When you configure a PBR policy, be careful to avoid this situation.

PBR tasks at a glance

To configure PBR, perform the following tasks:

1.     Configuring a policy

a.     Creating a node

b.     Setting match criteria for a node

c.     Configuring actions for a node

2.     Specifying a policy for PBR

Choose the following tasks as needed:

¡     Specifying a policy for local PBR

¡     Specifying a policy for interface PBR

Configuring a policy

Creating a node

1.     Enter system view.

system-view

2.     Create a node for a policy, and enter its view.

policy-based-route policy-name [ deny | permit ] node node-number

Setting match criteria for a node

1.     Enter system view.

system-view

2.     Enter policy node view.

policy-based-route policy-name [ deny | permit ] node node-number

3.     Set match criteria.

¡     Set an ACL match criterion.

if-match acl { acl-number | name acl-name }

By default, no ACL match criterion is set.

The ACL match criterion cannot match Layer 2 information.

¡     Set a packet length match criterion.

if-match packet-length min-len max-len

By default, no packet length match criterion is set.

Configuring actions for a node

About this task

The apply clauses allow you to specify the actions to be taken on matching packets on a node.

The following apply clauses determine the packet forwarding paths in a descending order:

·     apply access-vpn

·     apply next-hop

·     apply output-interface

·     apply default-next-hop

·     apply default-output-interface

PBR supports the apply clauses in Table 1.

Table 1 Apply clauses supported in PBR

Clause	Meaning	Remarks
apply precedence	Sets an IP precedence.	This clause is always executed.
apply ip-df df-value	Sets the Don't Fragment (DF) bit in the IP header.	This clause is always executed.
apply loadshare { next-hop | output-interface | default-next-hop | default-output-interface }	Enables load sharing among multiple next hops, output interfaces, default next hops, or default output interfaces.	Multiple next hop, output interface, default next hop, or default output interface options operate in either primary/backup or load sharing mode. ·     Primary/backup mode—One option is selected from all options in configuration order for packet forwarding, with all remaining options as backups. For example, if multiple output interfaces are configured, the first configured output interface is selected. When the selected output interface fails, the next available output interface takes over. ·     Load sharing mode—Matching traffic is distributed across the available options, as follows: ¡     Multiple output interface, default next hop, or default output interface options—Load share traffic in round robin manner, starting from the first configured option. They perform per-packet load sharing for traffic that does not match any fast forwarding entry, and perform per-flow load sharing for traffic that matches a fast forwarding entry. ¡     Multiple next hops—Load share traffic in proportion to their weight. By default, all next hops have the same weight and traffic is evenly distributed among them. By default, the primary/backup mode applies. For the load sharing mode to take effect, make sure multiple next hops, output interfaces, default next hops, or default output interfaces are set in the policy.
apply access-vpn	Specifies the forwarding tables that can be used for the matching packets.	Use this clause only in special scenarios that require sending packets received from one network to another network, for example, from a VPN to the public network, or from one VPN to another VPN. If a packet matches the forwarding table for a specified VPN instance, it is forwarded in the VPN.
apply remark-vpn	Enables VPN remark action.	VPN remark action marks the matching packets as belonging to the VPN instance to which they are forwarded based on the apply access-vpn vpn-instance command. All subsequent service modules of PBR handle the packets as belonging to the re-marked VPN instance. If the VPN remark action is not enabled, the forwarded matching packets are marked as belonging to the VPN instance or the public network from which they were received. VPN remark action applies only to packets that have been successfully forwarded based on the apply access-vpn vpn-instance command.
apply next-hop and apply output-interface	Sets next hops and sets output interfaces.	If both clauses are configured, only the apply next-hop clause is executed.
apply default-next-hop and apply default-output-interface	Sets default next hops and sets default output interfaces.	If both clauses are configured, only the apply default-next-hop clause is executed. The clauses take effect only in the following cases: ·     No next hops or output interfaces are set or the next hops and output interfaces are invalid. ·     The packet does not match any route in the routing table.
apply continue	Compares packets with the next node upon failure on the current node.	The apply continue clause applies when either of the the following conditions exist: ·     None of the following clauses is configured for packet forwarding: ¡     apply access-vpn vpn-instance ¡     apply next-hop ¡     apply output-interface ¡     apply default-next-hop ¡     apply default-output-interface ·     A clause listed above is configured, but it has become invalid. Then, a routing table lookup also fails for the matching packet. NOTE: A clause might become invalid because the specified next hop is unreachable, packets cannot be forwarded in the specified VPN instance, or the specified output interface is down.

‌

Restrictions and guidelines

For outbound PBR, you can specify only one next hop and the next hop must be directly connected.

If you specify a next hop or default next hop, PBR periodically performs a lookup in the FIB table to determine its availability. Temporary service interruption might occur if PBR does not update the route immediately after its availability status changes.

Configuring actions to modify packet fields

1.     Enter system view.

system-view

2.     Enter policy node view.

policy-based-route policy-name [ deny | permit ] node node-number

3.     Configure actions.

¡     Set an IP precedence.

apply precedence { type | value }

By default, no IP precedence is specified.

¡     Set the DF bit in the IP header.

apply ip-df df-value

By default, the DF bit in the IP header is not set.

Configuring actions to direct packet forwarding

1.     Enter system view.

system-view

2.     Enter policy node view.

policy-based-route policy-name [ deny | permit ] node node-number

3.     Configure actions.

¡     Specify the forwarding tables that can be used for the matching packets.

apply access-vpn { public | vpn-instance vpn-instance-name&<1-4> }

By default, the device forwards matching packets by using the forwarding table for the network from which the packets are received.

You can repeat this command to specify the forwarding tables for the public network and VPN instances. The device forwards the matching packets by using the first available forwarding table selected in the order in which they are specified.

¡     Enable VPN remark action to mark the matching packets as belonging to the VPN instance to which they are forwarded based on the apply access-vpn vpn-instance command.

apply remark-vpn

By default, VPN remark action is not configured.

¡     Set next hops.

apply next-hop [ vpn-instance vpn-instance-name | inbound-vpn ] { ip-address [ direct ] [ track track-entry-number ] [ weight weight-value ] }&<1-n>

By default, no next hops are specified.

On a node, you can specify a maximum of n next hops for backup or load sharing in one command line or by executing this command multiple times.

If multiple next hops on the same subnet are specified for backup, the device first uses the subnet route for the next hops to forward packets when the primary next hop fails. If the subnet route is not available, the device selects a backup next hop.

The value for n varies by device model. For more information, see the command reference.

¡     Enable load sharing among multiple next hops.

apply loadshare next-hop

By default, the next hops operate in primary/backup mode.

¡     Set output interfaces.

apply output-interface { interface-type interface-number [ track track-entry-number ] }&<1-4>

By default, no output interfaces are specified.

On a node, you can specify a maximum of four output interfaces for backup or load sharing in one command line or by executing this command multiple times.

¡     Enable load sharing among multiple output interfaces.

apply loadshare output-interface

By default, the output interfaces operate in primary/backup mode.

¡     Set default next hops.

apply default-next-hop [ vpn-instance vpn-instance-name | inbound-vpn ] { ip-address [ direct ] [ track track-entry-number ] }&<1-n>

By default, no default next hops are specified.

On a node, you can specify a maximum of n default next hops for backup or load sharing in one command line or by executing this command multiple times.

The value for n varies by device model. For more information, see the command reference.

¡     Enable load sharing among multiple default next hops.

apply loadshare default-next-hop

By default, the default next hops operate in primary/backup mode.

¡     Set default output interfaces.

apply default-output-interface { interface-type interface-number [ track track-entry-number ] }&<1-4>

By default, no default output interfaces are specified.

On a node, you can specify a maximum of four default output interfaces for backup or load sharing in one command line or by executing this command multiple times.

¡     Enable load sharing among multiple default output interfaces.

apply loadshare default-output-interface

By default, the default output interfaces operate in primary/backup mode.

Comparing packets with the next node upon match failure on the current node

1.     Enter system view.

system-view

2.     Enter policy node view.

policy-based-route policy-name [ deny | permit ] node node-number

3.     Compare packets with the next node upon match failure on the current node.

apply continue

By default, PBR does not compare packets with the next node upon match failure on the current node.

This command takes effect only when the match mode of the node is permit.

Specifying a policy for PBR

Specifying a policy for local PBR

About this task

Perform this task to specify a policy for local PBR to guide the forwarding of locally generated packets.

Restrictions and guidelines

You can specify only one policy for local PBR and must make sure the specified policy already exists. Before you apply a new policy, you must first remove the current policy.

Local PBR might affect local services such as ping and Telnet. When you use local PBR, make sure you fully understand its impact on local services of the device.

Procedure

1.     Enter system view.

system-view

2.     Specify a policy for local PBR.

ip local policy-based-route policy-name

By default, local PBR is not enabled.

Specifying a policy for interface PBR

About this task

Perform this task to apply a policy to an interface to guide the forwarding of packets received on the interface.

Restrictions and guidelines

You can apply only one policy to an interface and must make sure the specified policy already exists. Before you can apply a new interface PBR policy to an interface, you must first remove the current policy from the interface.

You can apply a policy to multiple interfaces.

Procedure

1.     Enter system view.

system-view

2.     Enter interface view.

interface interface-type interface-number

3.     Specify a policy for interface PBR.

ip policy-based-route policy-name

By default, no policy is applied to an interface.

Display and maintenance commands for PBR

Execute display commands in any view and reset commands in user view.

 

Task	Command
Display PBR policy information.	display ip policy-based-route [ policy policy-name ]
Display interface PBR configuration and statistics.	display ip policy-based-route interface interface-type interface-number [ slot slot-number ]
Display local PBR configuration and statistics.	display ip policy-based-route local [ slot slot-number ]
Display PBR configuration.	display ip policy-based-route setup
Clear PBR statistics.	reset ip policy-based-route statistics [ policy policy-name ]

‌

PBR configuration examples

Example: Configuring packet type-based local PBR

Network configuration

As shown in Figure 1, Device B and Device C do not have a route to reach each other.

Configure PBR on Device A to forward all TCP packets to the next hop 1.1.2.2 (Device B).

Figure 1 Network diagram

‌

Procedure

1.     Configure Device A:

# Configure the IP addresses of GigabitEthernet 1/0/1 and GigabitEthernet 1/0/2.

<DeviceA> system-view

[DeviceA] interface gigabitethernet 1/0/1

[DeviceA-GigabitEthernet1/0/1] ip address 1.1.2.1 24

[DeviceA-GigabitEthernet1/0/1] quit

[DeviceA] interface gigabitethernet 1/0/2

[DeviceA-GigabitEthernet1/0/2] ip address 1.1.3.1 24

[DeviceA-GigabitEthernet1/0/2] quit

# Configure ACL 3101 to match TCP packets.

[DeviceA] acl advanced 3101

[DeviceA-acl-ipv4-adv-3101] rule permit tcp

[DeviceA-acl-ipv4-adv-3101] quit

# Configure Node 5 for the policy aaa to forward TCP packets to next hop 1.1.2.2.

[DeviceA] policy-based-route aaa permit node 5

[DeviceA-pbr-aaa-5] if-match acl 3101

[DeviceA-pbr-aaa-5] apply next-hop 1.1.2.2

[DeviceA-pbr-aaa-5] quit

# Configure local PBR by applying the policy aaa to Device A.

[DeviceA] ip local policy-based-route aaa

2.     On Device B, configure the IP address of GigabitEthernet 1/0/1.

<DeviceB> system-view

[DeviceB] interface gigabitethernet 1/0/1

[DeviceB-GigabitEthernet1/0/1] ip address 1.1.2.2 24

3.     On Device C, configure the IP address of GigabitEthernet 1/0/2.

<DeviceC> system-view

[DeviceC] interface gigabitethernet 1/0/2

[DeviceC-GigabitEthernet1/0/2] ip address 1.1.3.2 24

Verifying the configuration

1.     Perform telnet operations to verify that local PBR on Device A operates as configured to forward the matching TCP packets to the next hop 1.1.2.2 (Device B), as follows:

# Verify that you can telnet to Device B from Device A successfully. (Details not shown.)

# Verify that you cannot telnet to Device C from Device A. (Details not shown.)

2.     Verify that Device A forwards packets other than TCP packets through GigabitEthernet 1/0/2. For example, verify that you can ping Device C from Device A. (Details not shown.)

Example: Configuring packet type-based interface PBR

Network configuration

As shown in Figure 2, Device B and Device C do not have a route to reach each other.

Configure PBR on Device A to forward all TCP packets received on GigabitEthernet 1/0/1 to the next hop 1.1.2.2 (Device B).

Figure 2 Network diagram

‌

Procedure

1.     Configure IP addresses and unicast routing protocol settings to make sure Device B and Device C can reach Host A. (Details not shown.)

2.     Configure Device A:

# Configure ACL 3101 to match TCP packets.

[DeviceA] acl advanced 3101

[DeviceA-acl-ipv4-adv-3101] rule permit tcp

[DeviceA-acl-ipv4-adv-3101] quit

# Configure Node 5 for the policy aaa to forward TCP packets to next hop 1.1.2.2.

[DeviceA] policy-based-route aaa permit node 5

[DeviceA-pbr-aaa-5] if-match acl 3101

[DeviceA-pbr-aaa-5] apply next-hop 1.1.2.2

[DeviceA-pbr-aaa-5] quit

# Configure interface PBR by applying the policy aaa to GigabitEthernet 1/0/1.

[DeviceA] interface gigabitethernet 1/0/1

[DeviceA-GigabitEthernet1/0/1] ip policy-based-route aaa

[DeviceA-GigabitEthernet1/0/1] quit

Verifying the configuration

1.     Perform telnet operations to verify that interface PBR on Device A operates as configured to forward the matching TCP packets to the next hop 1.1.2.2 (Device B), as follows:

# Verify that you can telnet to Device B from Host A successfully. (Details not shown.)

# Verify that you cannot telnet to Device C from Host A. (Details not shown.)

2.     Verify that Device A forwards packets other than TCP packets through GigabitEthernet 1/0/3. For example, verify that you can ping Device C from Host A. (Details not shown.)

Example: Configuring packet length-based interface PBR

Network configuration

As shown in Figure 3, configure interface PBR to guide the forwarding of packets received on GigabitEthernet 1/0/1 of Device A as follows:

·     Set the next hop of packets with a length of 64 to 300 bytes to 150.1.1.2/24.

·     Set the next hop of packets with a length of 301 to 1000 bytes to 151.1.1.2/24.

Figure 3 Network diagram

‌

Procedure

1.     Configure IP addresses and unicast routing protocol settings to make sure the devices each have a route to reach one another. (Details not shown.)

2.     Configure Device A:

# Configure Node 10 for the policy lab1 to forward packets with a length of 64 to 300 bytes to the next hop 150.1.1.2.

[DeviceA] policy-based-route lab1 permit node 10

[DeviceA-pbr-lab1-10] if-match packet-length 64 300

[DeviceA-pbr-lab1-10] apply next-hop 150.1.1.2

[DeviceA-pbr-lab1-10] quit

# Configure Node 20 for the policy lab1 to forward packets with a length of 301 to 1000 bytes to the next hop 151.1.1.2.

[DeviceA] policy-based-route lab1 permit node 20

[DeviceA-pbr-lab1-20] if-match packet-length 301 1000

[DeviceA-pbr-lab1-20] apply next-hop 151.1.1.2

[DeviceA-pbr-lab1-20] quit

# Configure interface PBR by applying the policy lab1 to GigabitEthernet 1/0/1.

[DeviceA] interface gigabitethernet 1/0/1

[DeviceA-GigabitEthernet1/0/1] ip policy-based-route lab1

[DeviceA-GigabitEthernet1/0/1] quit

Verifying the configuration

# Execute the debugging ip policy-based-route command on Device A.

<DeviceA> debugging ip policy-based-route

<DeviceA> terminal logging level 7

<DeviceA> terminal monitor

# Ping Loopback 0 of Device B from Host A, and set the data length to 64 bytes. In this way, the total length of the data packet is in the range of 64 to 300 bytes.

C:\>ping –n 1 -l 64 10.1.1.1

 

Pinging 10.1.1.1 with 64 bytes of data:

 

Reply from 10.1.1.1: bytes=64 time=1ms TTL=64

 

Ping statistics for 10.1.1.1:

    Packets: Sent = 1, Received = 1, Lost = 0 (0% loss),

Approximate round trip times in milli-seconds:

    Minimum = 1ms, Maximum = 1ms, Average = 1ms

The debugging information about PBR displayed on Device A is as follows:

<DeviceA>

*Jun  26 12:04:33:519 2012 DeviceA PBR4/7/PBR Forward Info: -MDC=1; Policy:lab1, Node:

 10,match succeeded.

*Jun  26 12:04:33:519 2012 DeviceA PBR4/7/PBR Forward Info: -MDC=1; apply next-hop 150

.1.1.2.

The output shows that Device A sets the next hop for the received packets to 150.1.1.2 according to PBR. The packets are forwarded through GigabitEthernet 1/0/2.

# Ping Loopback 0 of Device B from Host A, and set the data length to 300 bytes. In this way, the total length of the data packet is in the range of 301 to 1000 bytes.

C:\> ping –n 1 -l 300 10.1.1.1

 

Pinging 10.1.1.1 with 300 bytes of data:

 

Reply from 10.1.1.1: bytes=300 time=1ms TTL=64

 

Ping statistics for 10.1.1.1:

    Packets: Sent = 1, Received = 1, Lost = 0 (0% loss),

Approximate round trip times in milli-seconds:

    Minimum = 1ms, Maximum = 1ms, Average = 1ms

The debugging information about PBR displayed on Device A is as follows:

<DeviceA>

*Jun  26 12:20:33:610 2012 DeviceA PBR4/7/PBR Forward Info: -MDC=1; Policy:lab1, Node:

 20,match succeeded.

*Jun  26 12:20:33:610 2012 DeviceA PBR4/7/PBR Forward Info: -MDC=1; apply next-hop 151

.1.1.2.

The output shows that Device A sets the next hop for the received packets to 151.1.1.2 according to PBR. The packets are forwarded through GigabitEthernet 1/0/3.