Adjacent Node
Networking, explained. No BS.

MPLS

Printer friendly

What It Is

Multiprotocol Label Switching forwards packets using labels instead of only destination IP lookups. MPLS is still common in service provider cores, enterprise WANs, L3VPNs, pseudowires, traffic engineering, and segment routing designs. The modern view is broader than classic LDP: understand the MPLS data plane, then identify which control plane is installing labels.

Classic MPLS often uses IGP plus LDP. Modern transport may use SR-MPLS with IS-IS or OSPF extensions and no LDP.

Label Stack

Field Size Purpose
Label 20 bits Forwarding value
TC 3 bits Traffic class for QoS
S 1 bit Bottom of stack marker
TTL 8 bits Loop prevention and traceroute behavior

Modern note: MPLS can carry multiple labels. A packet may have a transport label, a VPN label, and service labels depending on the design.

Core Terms

Term Meaning
LSR Label switch router
PE Provider edge router, connects customer or service edge
P Provider core router, usually label switches only
CE Customer edge router
LSP Label-switched path
FEC Forwarding equivalence class
LIB Label information base
LFIB Label forwarding information base
PHP Penultimate hop popping
Explicit null Label used to preserve QoS or force final label lookup behavior

Planes

Plane Role Examples
Control plane Learns routes and exchanges labels IGP, LDP, RSVP-TE, BGP, IS-IS SR, OSPF SR
Data plane Forwards labeled packets Swap, push, pop
Service plane Adds VPN or service meaning L3VPN, EVPN, VPWS, VPLS

Watch out: MPLS forwarding can be healthy while the VPN service is broken, and the VPN can be healthy while the transport LSP is broken. Troubleshoot by layer.

Label Distribution Options

Method What It Does Modern Notes
LDP Distributes labels for IGP-chosen paths Still common in brownfield MPLS
RSVP-TE Signals traffic-engineered LSPs Less common in new designs
BGP labeled unicast Carries labels with BGP routes Used in inter-AS and some transport designs
Segment routing MPLS Encodes instructions as MPLS labels Reduces LDP/RSVP state, uses IGP extensions

Design note: SR-MPLS is still MPLS in the data plane. The difference is how labels are assigned and how paths are expressed.

Classic LDP Design

Requirement Why It Matters
IGP reachability LDP follows routed next hops
Stable loopbacks LDP router IDs and BGP next hops need reachability
MPLS on core links Labels must exist hop by hop
MTU headroom Label stacks add overhead
LDP-IGP sync Avoid blackholing during convergence
Route consistency LFIB depends on FIB and labels

Watch out: If the IGP chooses a path where MPLS is not enabled on every hop, labeled traffic can fail even though IP ping works.

Cisco IOS/IOS-XE Examples

Classic LDP core:

ip cef
!
mpls label protocol ldp
mpls ldp router-id Loopback0 force
!
interface Loopback0
 ip address 10.255.0.1 255.255.255.255
!
interface GigabitEthernet0/0
 description MPLS core link to P2
 ip address 10.0.12.1 255.255.255.252
 mpls ip
 mpls mtu 1512

OSPF LDP autoconfig pattern:

router ospf 100
 mpls ldp autoconfig area 0

Segment routing MPLS pattern:

segment-routing mpls
 global-block 16000 23999
!
router isis CORE
 address-family ipv4 unicast
  segment-routing mpls

Notes:

  • MPLS, LDP, SR-MPLS, and VPN feature support varies heavily by platform license and ASIC.
  • Do not enable MPLS toward access or untrusted networks.
  • Raise MTU only where the underlay supports it end to end.

Troubleshooting

Symptom Check Likely Cause
No LDP neighbor MPLS enabled, LDP discovery, TCP/646, IGP LDP not forming
Label missing LIB, LFIB, route to FEC No binding or no route
IP works, VPN fails MP-BGP VPNv4/VPNv6, VRF routes, labels Service label issue
Labeled ping fails MPLS MTU, label path, LFIB Transport LSP issue
Traceroute oddity PHP, explicit null, TTL propagation Normal MPLS behavior or policy
Blackhole after failure LDP-IGP sync, convergence, stale labels IGP and LDP out of sync
QoS lost TC mapping, explicit null, provider policy QoS not preserved across label stack

Commands

show mpls interfaces
show mpls ldp neighbor
show mpls ldp bindings
show mpls forwarding-table
show ip cef <prefix> detail
show mpls ldp discovery
show bgp vpnv4 unicast all labels
show segment-routing mpls connected-prefix-sid-map
traceroute mpls ipv4 <prefix> <mask>

Expected clues:

  • Core-facing interfaces show MPLS enabled.
  • LDP neighbors match the intended core topology.
  • LFIB has labels for loopbacks and transport prefixes.
  • VPN routes have service labels where L3VPN is used.
  • MTU supports the expected label stack.

Watch Out

  • Do not confuse IP reachability with labeled reachability.
  • Do not ignore MTU. MPLS labels add 4 bytes each.
  • Do not enable MPLS on customer-facing interfaces unless the service requires it.
  • Do not migrate from LDP to segment routing without planning interworking.
  • Do not troubleshoot L3VPN before proving the transport LSP is healthy.

References