Configuring AToM

Simple misconfiguration is a common cause of issues with AToM, so this section focuses on configuration of PE routers in an AToM environment. Configuration of provider (P) routers is identical to that discussed in the section "Configuring the P Router" in Chapter 6.

Note that it may be useful to reference Figure 7-14 as you read this section, although not all attachment circuit types are illustrated.

Figure 7-14. AToM Reference Topology

[View full size image]

Configuration of the PE routers consists of seven steps:

The sections that follow provide detailed information on each of these steps.

Step 1: Configure the Loopback Interface to Be Used as the LDP Router ID

Example 7-1 shows the configuration of the loopback interface is shown in Example 7-1.

Example 7-1. Configuration of the Loopback Interface

interface Loopback0
 ip address 172.16.1.1 255.255.255.255

Note that the IP address configured on the loopback interface should have a 32-bit mask; otherwise, AToM may not function correctly.

Step 2: Enable CEF

CEF must be enabled on the PE router or MPLS will not function.

Example 7-2 shows the configuration of CEF.

Example 7-2. Configuration of CEF

ip cef [distributed]

The distributed keyword shown in Example 7-2 is used to enable distributed CEF (dCEF). Distributed CEF is available on high-end platforms such as the 7500 and 12000 series routers.

Step 3: Specify LDP as the Label Protocol

LDP should now be specified as the label protocol as demonstrated in Example 7-3.

Example 7-3. LDP Is Specified as the Label Distribution Protocol

mpls label protocol ldp

In Example 7-3, the mpls label protocol ldp command specifies LDP as the label distribution protocol.

Step 4: Configure the LDP Router ID

Once LDP has been specified as the label protocol, the next recommended step is to configure the LDP router ID.

Example 7-4 shows the configuration of the LDP router ID.

Example 7-4. Configuration of the LDP Router ID

mpls ldp router-id Loopback0 force

In Example 7-4, the loopback interface address configured in Step 1 is specified as the LDP router ID. The force keyword ensures that it becomes the LDP router ID.

Step 5: Configure MPLS on Core Interfaces

The mpls ip command is used to enable MPLS on core (frame mode) interfaces, as demonstrated in Example 7-5.

Example 7-5. Configuration of MPLS on Core (Frame Mode) Interfaces

interface FastEthernet2/0
 mpls ip

If LSRs are connected via an ATM PVC, the interface should be configured as shown in Example 7-6.

Example 7-6. Configuration of MPLS over an ATM PVC (Frame Mode)

interface ATM3/0.1 point-to-point
 ip address 172.16.7.1 255.255.255.0
 pvc 1/50
  encapsulation aal5snap
 !
 mpls ip

MPLS is enabled on an ATM PVC with VPI/VCI 1/50 in Example 7-6. Note that the subinterface type is point-to-point.

If LSRs are connected via label switching controlled ATM (LC-ATM) interfaces, then the interface configuration is as shown in Example 7-7.

Example 7-7. Configuration of an LC-ATM Interface (Cell Mode)

interface ATM3/0.1 mpls
 ip address 172.16.7.1 255.255.255.0
 mpls ip

As you can see, the subinterface type specified in Example 7-7 is mpls. Note that the mpls ip command is also configured on the interface itself.

Step 6: Configure the MPLS Backbone IGP

The MPLS backbone Interior Gateway Protocol (IGP) should now be configured.

The two most commonly used IGPs within the MPLS backbone are Intermediate System-to-Intermediate System (IS-IS) and Open Shortest Path First (OSPF). These are often used because they are currently the only two IGPs that support MPLS traffic engineering. This section contains sample configurations for both IS-IS and OSPF.

Configuring IS-IS as the MPLS Backbone IGP

Example 7-8 shows configuration of IS-IS for IP reachability across the MPLS backbone.

Example 7-8. Configuration of IS-IS for IP Reachability

router isis
 passive-interface Loopback0
 net 49.0001.0000.0000.0001.00
 is-type level-2-only
 metric-style wide

The router isis command enables IS-IS on the PE router.

Interface loopback 0 (configured in Step 1) is then enabled for IS-IS using the passive-interface loopback0 command.

Make sure that the loopback interface used as the LDP router ID is enabled for IS-IS using the passive-interface command. If the loopback interface is not advertised in IS-IS, LDP discovery and session establishment will fail. Note that some versions of IOS also require you to configure the ip router isis command on the loopback interface in addition to specifying the passive interface command

The net 49.0001.0000.0000.0001.00 command is used to configure the network entity title (NET). This consists of the area ID (49.0001), the system ID (0000.0000.0001), and the selector (00).

The is-type level-2-only command configures the PE router as a backbone router.

Finally, the PE router is configured to support 24- or 32-bit metrics using the metric-style wide command. Wide metrics are required if you are using MPLS traffic engineering within the backbone.

After globally enabling IS-IS on the PE router, the next step is to configure IS-IS on core interfaces using the ip router isis command, as demonstrated in Example 7-9.

Example 7-9. Configuration of IS-IS on Core Interfaces

interface FastEthernet2/0
 ip router isis

In Example 7-9, IS-IS is enabled on core interface fast Ethernet 2/0 using the ip router isis command.

Configuring OSPF as the MPLS Backbone IGP

As previously mentioned, the other IGP that is commonly used within an MPLS backbone is OSPF.

Example 7-10 shows configuration of OSPF for IP reachability within the MPLS backbone.

Example 7-10. Configuration of OSPF for IP Reachability

router ospf 100
 passive-interface Loopback0
 network 172.16.0.0 0.0.255.255 area 0

The router ospf 100 command enables OSPF process 100 on the PE router.

The passive-interface loopback0 command is used to ensure that OSPF packets are not transmitted on interface loopback 0.

Finally, interface loopback 0, together with core facing interfaces (in this case, fast Ethernet 2/0) are placed in area 0 using the network 172.16.0.0 0.0.255.255 area 0 command.

Step 7: Configure the AToM Pseudowires

The next step is to configure the AToM pseudowires. Configuration depends on the encapsulation type used on the attachment circuits.

Configuration for Ethernet, HDLC, PPP, ATM (AAL5 and cell relay), and Frame Relay encapsulation types is detailed in the sections that follow.

When configuring an AToM pseudowire, ensure that the VC ID is the same on both PE routers.

Ethernet Encapsulation

There are two modes of transport for Ethernet PDUs over an AToM pseudowire:

• Port mode The entire Ethernet frame (minus the preamble and FCS) is transported over the pseudowire.

• VLAN (802.1Q) mode Traffic from an 802.1q VLAN is transported over the pseudowire.

Ethernet Port Mode

An Ethernet port mode attachment circuit should be configured as shown in Example 7-11.

Example 7-11. Configuration of an Ethernet Port Mode Attachment Circuit

interface FastEthernet0/0
 xconnect 172.16.1.4 121 encapsulation mpls

In Example 7-11, interface fast Ethernet 0/0 is bound to the pseudowire with VCID 121 to peer PE router 172.16.1.4 using the xconnect peer_LDP_router_id VCID encapsulation mpls command.

Ethernet VLAN (802.1Q) Mode

If the encapsulation type used on the attachment circuit is Ethernet VLAN (802.1q), configuration is as shown in Example 7-12.

Example 7-12. Configuration of an Ethernet VLAN Mode (802.1Q) Attachment Circuit

interface FastEthernet1/0
!
interface FastEthernet1/0.1
 encapsulation dot1Q 10
 xconnect 172.16.1.4 121 encapsulation mpls

In highlighted line 1, 802.1Q encapsulation for VLAN ID 10 is configured on interface fast Ethernet 1/0.1 using the encapsulation dot1q vlan_id command.

The xconnect peer_LDP_router_id VCID encapsulation mpls command is then used to bind interface fast Ethernet 1/0.1 to the AToM pseudowire to peer PE router 172.16.1.4 with VC ID 121.

Frame Relay Encapsulation

A pseudowire used to transport Frame Relay can be configured in two ways:

• A DLCI-to-DLCI connection When a DLCI-to-DLCI connection is used, the PE router functions as a Frame Relay switch and participates in LMI.

• A port-to-port connection When a port-to-port connection is used, Frame Relay frames are simply transported transparently over the pseudowire. In this case, the PE router does not participate in LMI.

Frame Relay DLCI-to-DLCI Connections

Example 7-13 shows the configuration of a DLCI-to-DLCI connection.

Example 7-13. Configuration of a DLCI-to-DLCI Connection

frame-relay switching
interface Serial4/0
 encapsulation frame-relay
 frame-relay intf-type dce

The frame-relay switching global configuration mode command is used to enable Frame Relay permanent virtual circuit (PVC) switching.

The encapsulation frame-relay [ietf] command is used to configure Frame Relay encapsulation on the attachment circuit interface. Note that the default encapsulation type is Cisco, and the ietf keyword enables IETF RFC 1490 encapsulation.

The interface type is then configured using the frame-relay intf-type [dce | dte | nni] command. In this case, the interface type is User-to-Network Interface (UNI) data circuit-terminating equipment (DCE). The dte and nni keywords can be used to configure the interface as UNI data terminal equipment (DTE) and Network-to-Network Interface, respectively.

The next step is to bind the attachment circuit to the pseudowire, as shown in Example 7-14.

Example 7-14. Binding the PVC to the Pseudowire

connect PVCtoParis Serial4/0 50 l2transport
 xconnect 172.16.1.4 123 encapsulation mpls

The connect connection_name interface dlci l2transport command is used to specify a locally significant connection name, attachment circuit interface, and DLCI.

The xconnect peer_LDP_router_ID VCID encapsulation mpls command is then used to bind the PVC to the specified VC and peer PE router address. In Example 7-14, DLCI 50 (on interface serial 4/0) is bound to a pseudowire with VC ID 123 to peer 172.16.1.4.

It worth noting that it is possible to use Cisco encapsulation at one end and IETF encapsulation at the other end of the same pseudowire. Similarly, different LMI types (Cisco/ANSI/Q933a) may be used at either end of the same pseudowire.

Frame Relay Port-to-Port Connections

Example 7-15 shows the configuration of a Frame Relay port-to-port connection.

Example 7-15. Configuration of a Frame Relay Port-to-Port Connection

interface Serial4/2
 xconnect 172.16.1.4 125 encapsulation mpls

In Example 7-15, interface serial 4/2 is bound to a pseudowire with VC ID 125 to peer PE router 172.16.1.4. Note that the encapsulation type specified on interface serial 4/2 is HDLC (the default) and not Frame Relay.

AAL5 Encapsulation

Example 7-16 shows the configuration of AAL5 over MPLS.

Example 7-16. Configuration of AAL5 over MPLS

interface ATM3/0
 no ip address
 no ip directed-broadcast
 no atm enable-ilmi-trap
 no atm ilmi-keepalive
!
interface ATM3/0.1 point-to-point
 no ip directed-broadcast
 no atm enable-ilmi-trap
 pvc 1/50 l2transport
  encapsulation aal5
  xconnect 172.16.1.4 122 encapsulation mpls

The pvc vpi/vci l2transport command indicates that the PVC with the specified virtual path indicator (VPI) and virtual circuit indicator (VCI) will be bound to a pseudowire.

The encapsulation aal5 command configures AAL5 encapsulation for the interface.

Finally, the xconnect peer_LDP_router_ID VCID encapsulation mpls command binds the PVC to the pseudowire with the specified VC ID to the peer PE router.

In Example 7-16, the PVC with VCI/VPI 1/50 is bound to the pseudowire with VC ID 122 to peer PE router 172.16.1.4.

ATM Cell Relay

ATM cell relay can be configured in various modes:

• Virtual circuit (VC) mode

• Virtual path (VP) mode

• Port mode

It is also possible to configure single cell relay, in which one ATM cell is transported in each MPLS packet, and packed (concatenated) cell relay, in which multiple cells are transported in each MPLS packet.

ATM Cell Relay: VC Mode

In VC mode, cell relay is configured for a PVC. Example 7-17 shows a sample configuration on a point-to-point subinterface.

Example 7-17. Configuration of ATM Cell Relay: VC Mode

interface ATM3/0
 no ip address
 no ip directed-broadcast
 atm mcpt-timers 250 500 750
 no atm enable-ilmi-trap
 no atm ilmi-keepalive
!
interface ATM3/0.1 point-to-point
 no ip directed-broadcast
 no atm enable-ilmi-trap
 pvc 1/50 l2transport
  encapsulation aal0
  cell-packing 10 mcpt-timer 1
  xconnect 172.16.1.4 122 encapsulation mpls

In highlighted line 2, the pvc vpi/vci l2transport command is used to indicate that the PVC with the specified VPI/VCI will be bound to a pseudowire.

The encapsulation aal0 command in highlighted line 3 is then used to specify ATM adaptation layer 0 (AAL0) (cell relay).

The xconnect peer_LDP_router_id vcid encapsulation mpls command in highlighted line 5 is used to bind the interface to the pseudowire with VC ID 122 to peer PE router 172.16.1.4.

Note also the optional commands in highlighted lines 1 and 4: atm mcpt-timers 250 500 750 and cell-packing 10 mcpt-timer 1. These two commands are used to enable cell packing (single cell relay is used by default).

The atm mcpt-timers timer1-timeout1 timer2-timeout2 timer3-timeout3 command is used to specify maximum cell packing timeouts (MCPT), which control how long the PE router waits for ATM cells on the attachment circuit before packing them into an MPLS packet and transmitting the packet across the backbone. In this example, timeouts of 250, 500, and 750 microseconds are specified.

The cell-packing cells mcpt-timer timer command is used to enable cell packing and specify the maximum number of cells that the router will in pack into an MPLS packet, as well as the MCPT to use. In this example, a maximum of 10 cells and an MCPT of 1 (250 microseconds) are specified.

Note that if the timer specified using the cell-packing cells mcpt-timer timer command is reached before the maximum number of cells has arrived, the packet is transmitted without waiting for more cells.

ATM Cell Relay: VP Mode

If you configure VP mode, cell relay is enabled for a permanent virtual path (PVP), as demonstrated in Example 7-18.

Example 7-18. Configuration of ATM Cell Relay: VP Mode

interface ATM3/0
 no ip address
 no ip directed-broadcast
 atm mcpt-timers 250 500 750
 atm pvp 10 l2transport
  cell-packing 10 mcpt-timer 1
  xconnect 172.16.1.4 122 encapsulation mpls

The atm pvp vpi l2transport command in highlighted line 2 is used to indicate that the PVP with the specified VPI will be bound to a pseudowire.

In highlighted line 4, the xconnect peer_LDP_router_id vcid encapsulation mpls command is used to bind the PVP to a pseudowire with VC ID 122 to peer PE router 172.16.1.4.

Again, the optional atm mcpt-timers timer1-timeout1 timer2-timeout2 timer3-timeout3 and cell-packing cells mcpt-timer timer commands in highlighted lines 1 and 3 are used to configure MCPT timers and enable cell packing. Note that cell packing is currently supported for only VC and VP mode cell relay.

ATM Cell Relay: Port Mode

Port mode, currently supported on only 12000 series routers, allows a single ATM cell arriving on the main ATM interface to be packed into an MPLS packet.

Example 7-19 demonstrates configuration for port mode.

Example 7-19. Configuration of ATM Cell Relay: Port Mode

interface ATM3/0
 xconnect 172.16.1.4 122 encapsulation mpls

In Example 7-19, the xconnect peer_LDP_router_id vcid encapsulation mpls command is used to bind the ATM interface to a pseudowire with VCID 122 to peer 172.16.1.4.

HDLC Encapsulation

Example 7-20 shows the configuration of HDLC over MPLS.

Example 7-20. Configuration of HDLC over MPLS

interface Serial4/1
 xconnect 172.16.1.4 124 encapsulation mpls

In Example 7-20, the xconnect peer_LDP_router_id vcid encapsulation mpls command is used to bind interface serial 4/1 to a pseudowire with VC ID 124 to peer PE router 172.16.1.4.

Note that HDLC is the default encapsulation type, so the command encapsulation hdlc is not shown in Example 7-20.

PPP Encapsulation

Example 7-21 shows configuration of PPP over MPLS.

Example 7-21. Configuration of PPP over MPLS

interface Serial4/2
 encapsulation ppp
 xconnect 172.16.1.4 125 encapsulation mpls

The encapsulation ppp command is used to configure PPP encapsulation on interface serial 4/2.

The xconnect peer_LDP_router_id vcid encapsulation mpls command is then used to bind the interface to a pseudowire with VC ID 125 to peer PE router 172.16.1.4.

That concludes the configuration of the PE router.

Complete Sample Configurations for AToM PE Routers

Example 7-22 shows complete sample configurations for peer PE routers London_PE and Paris_PE.

Example 7-22. Sample AToM PE Router Configurations

! London_PE:
London_PE#show running-config
Building configuration...
Current configuration : 2249 bytes
!
version 12.0
service nagle
no service pad
service tcp-keepalives-in
service timestamps debug datetime msec localtime show-timezone
service timestamps log datetime msec localtime show-timezone
service password-encryption
!
hostname London_PE
!
logging buffered 16384 debugging
no logging console
enable secret 5 $1$roOI$Lgw4QtRnK/93LeNAkJYAi0
!
ip subnet-zero
no ip source-route
! Enable Cisco Express Forwarding (CEF)
ip cef
!
!
no ip finger
no ip bootp server
!
! Configure frame-relay switching (for DLCI-to-DCLI switching)
frame-relay switching
!
! Configure the label protocol (LDP)
mpls label protocol ldp
no mpls traffic-eng auto-bw timers frequency 0
!
! Configure the LDP router ID (= mpls ldp router-id Loopback0 force)
tag-switching tdp router-id Loopback0 force
!
! Configure the loopback interface
interface Loopback0
 ip address 172.16.1.1 255.255.255.255
 no ip directed-broadcast
!
interface FastEthernet1/0
 no ip address
 no ip directed-broadcast
!
! Configure a VLAN (802.1q) attachment circuit (and pseudowire to Paris_PE)
interface FastEthernet1/0.1
 encapsulation dot1Q 10
 no ip directed-broadcast
 xconnect 172.16.1.4 121 encapsulation mpls
 no cdp enable
!
! Configure MPLS on core interfaces (tag-switching ip = mpls ip)
interface FastEthernet2/0
 ip address 172.16.2.1 255.255.255.0
 no ip redirects
 no ip directed-broadcast
 no ip proxy-arp
 ip router isis
 tag-switching ip
 no cdp enable
!
! Configure Frame-Relay DLCI-to-DLCI switching
interface Serial4/0
 no ip address
 no ip directed-broadcast
 encapsulation frame-relay
 no ip mroute-cache
 no fair-queue
 frame-relay intf-type dce
!
! Configure an HDLC pseudowire to Paris_PE
interface Serial4/1
 no ip address
 no ip directed-broadcast
 xconnect 172.16.1.4 124 encapsulation mpls
 no cdp enable
!
!
! Configure the MPLS backbone IGP
router isis
 passive-interface Loopback0
 net 49.0001.0000.0000.0001.00
 is-type level-2-only
 metric-style wide
!
ip classless
!
logging trap debugging
!
! Bind DLCI 50 to a pseudowire (to Paris_PE)
connect PVCtoParis Serial4/0 50 l2transport
 xconnect 172.16.1.4 123 encapsulation mpls
 !
line con 0
 exec-timeout 0 0
 password 7 01100F175804
 login
line aux 0
line vty 0 4
 password 7 094F471A1A0A
 login
!
end
________________________________________________________________
! Paris_PE:
Paris_PE#show running-config
Building configuration...

Current configuration : 1893 bytes
!
version 12.0
service nagle
no service pad
service tcp-keepalives-in
service timestamps debug datetime msec localtime show-timezone
service timestamps log datetime msec localtime show-timezone
service password-encryption
!
hostname Paris_PE
!
logging buffered 16384 debugging
no logging console
enable secret 5 $1$KwQz$V6quL3xhdKp7yUNosh2/7/
!
ip subnet-zero
no ip source-route
!
!
! Enable Cisco Express Forwarding (CEF)
ip cef
no ip finger
no ip bootp server
!
! Configure frame-relay switching (for DLCI-to-DCLI switching)
frame-relay switching
!
! Configure the label protocol (LDP)
mpls label protocol ldp
no mpls traffic-eng auto-bw timers frequency 0
!
! Configure the LDP router ID (= mpls ldp router-id Loopback0 force)
tag-switching tdp router-id Loopback0 force
!
! Configure the loopback interface
interface Loopback0
 ip address 172.16.1.4 255.255.255.255
 no ip directed-broadcast
!
interface FastEthernet0/0
 no ip address
 no ip directed-broadcast
!
! Configure a VLAN (802.1q) attachment circuit (and pseudowire to London_PE)
interface FastEthernet0/0.1
 encapsulation dot1Q 10
 no ip directed-broadcast
 no cdp enable
 xconnect 172.16.1.1 121 encapsulation mpls
!
! Configure MPLS on core interfaces (tag-switching ip = mpls ip)
interface FastEthernet1/0
 ip address 172.16.4.2 255.255.255.0
 no ip redirects
 no ip directed-broadcast
 no ip proxy-arp
 ip router isis
 tag-switching ip
 no cdp enable
!
! Configure Frame-Relay DLCI-to-DLCI switching
interface Serial2/0
 no ip address
 no ip directed-broadcast
 encapsulation frame-relay
 no ip mroute-cache
 no fair-queue
 frame-relay intf-type dce
!
! Configure an HDLC pseudowire to London_PE
interface Serial2/2
 no ip address
 no ip directed-broadcast
 no cdp enable
 xconnect 172.16.1.1 124 encapsulation mpls
!
! Configure the MPLS backbone IGP
router isis
 net 49.0001.0000.0000.0004.00
 is-type level-2-only
 metric-style wide
 passive-interface Loopback0
!
ip classless
!
! Bind DLCI 60 to a pseudowire (to London_PE)
connect PVCtoBank Serial2/0 60 l2transport
 xconnect 172.16.1.1 123 encapsulation mpls
 !
!
line con 0
 exec-timeout 0 0
 password 7 030752180500
 login
line aux 0
Sline vty 0 4
 privilege level 15
 password 7 121A0C041104
 login
!
end

In Example 7-22, three pseudowires are configured between peer PE routers London_PE (172.16.1.1) and Paris_PE (172.16.1.4):

• One Ethernet VLAN (802.1Q) pseudowire

• One Frame Relay DLCI-to-DLCI pseudowire

• One HDLC pseudowire

Note that some commands discussed in this section, such as mpls ip, are not readily apparent in the configurations shown in Example 7-22. This is because the mpls keyword is translated into the tag-switching keyword for backward compatibility with Cisco IOS versions that do not support the mpls keyword.

Maximum Transmission Unit Issues

When configuring AToM, it is important to be aware of possible maximum transmission unit (MTU) issues.

If the size of packets transmitted over the AToM pseudowire (see Figure 7-2) is larger than the MPLS MTU of any of the links between PE routers across the MPLS backbone, then these packets will be dropped.

In this scenario, large packets are not being successfully transmitted across the HDLC pseudowire from router cisco_CE1 to cisco_CE2 (see Figure 7-14). The output of the extended ping in Example 7-23 illustrates the transmission failure of large packets across the pseudowire.

Example 7-23. Large Packets Are Not Successfully Transmitted Across the Pseudowire

cisco_CE1#ping
Protocol [ip]:
Target IP address: 192.168.1.2
Repeat count [5]:
Datagram size [100]: 1500
Timeout in seconds [2]:
Extended commands [n]:
Sweep range of sizes [n]:
Type escape sequence to abort.
Sending 5, 1500-byte ICMP Echos to 192.168.1.2, timeout is 2 seconds:
.....
Success rate is 0 percent (0/5)
cisco_CE1#

In Example 7-23, a 1500 byte ping is sent from router cisco_CE1 to cisco_CE2 (192.168.1.2). As highlighted line 3 shows, the ping is not successful.

The output of the debug mpls l2transport packet error command in Example 7-24 reveals the problem.

Example 7-24. debug mpls l2transport packet error Command Output

London_PE#debug mpls l2transport packet error
AToM packet errors debugging is on
London_PE#
*Jan 26 19:28:18.507 UTC: ATOM-HDLC imposition: in Se4/1, circuit id 7,
  size 1508, packet dropped, output MTU exceeded
*Jan 26 19:28:20.507 UTC: ATOM-HDLC imposition: in Se4/1, circuit id 7,
  size 1508, packet dropped, output MTU exceeded
London_PE#

In highlighted lines 1 and 2, you can see that the ping packets received by London_PE from cisco_CE1 are being dropped as they are switched across the HDLC pseudowire to Paris_PE. This is because the AToM packet size (1508 bytes) exceeds the output MTU (the MTU of interface Fast Ethernet 2/0 on London_PE). Note that the AToM packet size shown here excludes the tunnel and VC labels.

The MPLS MTU on interface Fast Ethernet is then verified using the show mpls interfaces interface detail command, as demonstrated in Example 7-25.

Example 7-25. Verifying the MTU on Interface Fast Ethernet 2/0

London_PE#show mpls interfaces FastEthernet 2/0 detail
Interface FastEthernet2/0:
        IP labeling enabled (ldp)
        LSP Tunnel labeling not enabled
        BGP labeling not enabled
        MPLS operational
        Optimum Switching Vectors:
          IP to MPLS Turbo Vector
          MPLS Turbo Vector
        Fast Switching Vectors:
          IP to MPLS Fast Switching Vector
          MPLS Turbo Vector
        MTU = 1500
London_PE#

As you can see, the MPLS MTU is 1500 bytes on interface Fast Ethernet 2/0. Far too small!

To ensure that large packets are not dropped, make sure that the MPLS MTU is correctly configured on P and core facing PE router interfaces.

The MPLS MTU on core (backbone) links should be greater than or equal to the following:

label stack (number of labels * 4 bytes)

+ control word (4 bytes)

+ Layer 2 PDU (MTU of attachment circuit + attachment circuit Layer 2 header)

You can verify the presence of the control word (cbit=0 when it is not present, and cbit=1 when it is) on a pseudowire (VC), as well as the attachment circuit MTU using the show mpls l2transport binding command. See the section entitled "Other AToM Troubleshooting Commands" toward the end of this chapter for more information on this command.

If you are transporting HDLC PDUs (attachment circuit MTU of 1500 bytes and attachment circuit Layer 2 header of 4 bytes), and the label stack consists of two labels (each of 4 bytes), the required backbone MTU between the PE routers is as follows:

8 (label stack1 tunnel label + VC label)

+ 4 (control word)

+ 1504 (Layer 2 PDU1500 byte attachment circuit MTU + 4 byte HDLC header)

= 1516 bytes

This MTU (or greater) should be configured on all P and PE core interfaces, using the mtu bytes or mpls mtu bytes command, as shown in Example 7-26. The mpls mtu bytes command is required for some interface types. In fact, the MTU configured on core interfaces should accommodate all pseudowires between PE routers. Although in this example, only the HDLC pseudowire is being considered, in reality, you would also have to consider the Ethernet VLAN and Frame Relay DLCI to DLCI pseudowires between London_PE and Paris_PE.

Example 7-26. Adjusting the MPLS MTU on London_PE, Interface Fast Ethernet 2/0

London_PE#conf t
Enter configuration commands, one per line.  End with CNTL/Z.
London_PE(config)#interface FastEthernet 2/0
London_PE(config-if)#mpls mtu 1516
London_PE(config-if)#end
London_PE#

Example 7-27 shows that once the MPLS MTU has been modified, large packets are successfully transmitted across the HDLC pseudowire.

Example 7-27. Large Packets Are Now Successfully Transmitted Across the HDLC Pseudowire

cisco_CE1#ping
Protocol [ip]:
Target IP address: 192.168.1.2
Repeat count [5]:
Datagram size [100]: 1500
Timeout in seconds [2]:
Extended commands [n]:
Sweep range of sizes [n]:
Type escape sequence to abort.
Sending 5, 1500-byte ICMP Echos to 192.168.1.2, timeout is 2 seconds:
!!!!!
Success rate is 100 percent (5/5), round-trip min/avg/max = 789/790/794 ms
cisco_CE1#

Note that the label stack can be of variable size, as the following list explains:

• If the PE routers are in a back-to-back configuration, the label stack consists of one label, the VC label.

• If you are using LDP to signal the tunnel LSP, the label stack consists of two labels, the VC label and the IGP (LDP) label.

• If you are additionally using traffic engineering tunnels between P routers, the label stack consists of three labels, the VC label, the IGP label, and the TE label.

• If you are using fast reroute with traffic engineering tunnels between P routers, the label stack consists of four labels, the VC label, the IGP label, the TE label, and the fast reroute label.

If there are any Ethernet switches in the Points-of-Presence (POPs), also ensure that jumbo frame support is configured on these switches.