Migrating from LDP to Segment Routing is one of the most common infrastructure projects facing service provider network teams. The good news: OcNOS supports a phased migration model where LDP and SR-MPLS coexist on the same network simultaneously — allowing operators to migrate one node at a time without any service disruption or maintenance windows.
Why Migrate from LDP to SR?
| Capacité | LDP | SR-MPLS |
|---|---|---|
| Signaling protocol | Separate LDP sessions | None — IGP distributes SIDs |
| Transit router state | Per-prefix label binding | Stateless — only node SIDs |
| Traffic engineering | Not supported | SR-TE + Flex-Algo |
| Reroutage rapide | LFA only (~60–80%) | TI-LFA (100%) |
| IGP synchronization | Complex (LDP-IGP sync) | Native — no sync needed |
| Operational complexity | High (two protocol stacks) | Low (IGP only) |
Migration Strategy: Three Phases
Phase 1: Enable SR with LDP Coexistence
In Phase 1, Segment Routing is enabled on the IS-IS domain while LDP continues to operate. Both protocols distribute labels simultaneously. Traffic continues to use LDP labels — this phase is purely additive and carries zero risk.
! OcNOS Phase 1 -- Enable IS-IS SR alongside existing LDP
!
! Step 1: Enable SR globally under IS-IS (non-disruptive)
router isis CORE
segment-routing mpls
!
address-family ipv4 unicast
segment-routing mpls
exit-address-family
!
! Step 2: Assign Node SID to loopback (new config, no traffic impact)
interface lo
ip address 10.0.0.1/32
ip router isis CORE
isis segment-routing prefix-sid index 1
!
! Step 3: LDP remains fully operational -- no changes needed
! Both LDP and SR labels are now distributed
!
! Verify SR is active alongside LDP:
show isis segment-routing prefix-sids
show mpls ldp neighbor ! LDP still active
show mpls forwarding-table ! Both LDP and SR labels present
Phase 2: Prefer SR Labels (Traffic Cutover)
Phase 2 switches traffic from LDP labels to SR labels using the sr-prefer knob under IS-IS. LDP sessions remain active as a fallback during validation. This is the actual traffic migration step — it can be applied per-node and rolled back immediately if needed.
! OcNOS Phase 2 -- Prefer SR labels, keep LDP as fallback
!
router isis CORE
segment-routing mpls
segment-routing sr-prefer ! Prefer SR labels over LDP labels
!
address-family ipv4 unicast
segment-routing mpls
fast-reroute per-prefix ti-lfa ! Optionally enable TI-LFA now
exit-address-family
!
! Verify traffic is using SR labels (LDP labels still installed as backup):
show mpls forwarding-table
! Look for: primary label = SR label, backup = LDP label
!
! Verify MPLS path for a specific destination:
show ip route 10.0.0.5
show mpls forwarding-table 10.0.0.5
!
! Check LDP is still operational (fallback):
show mpls ldp neighbor
show mpls ldp bindings
Phase 3: Decommission LDP
Once all nodes are running SR-prefer and traffic has been validated on SR labels, LDP can be removed. This is the final cleanup step.
! OcNOS Phase 3 -- Remove LDP, SR-only operation
!
! Step 1: Remove LDP configuration
no mpls ldp
!
! Step 2: Remove LDP-IGP sync if configured
router isis CORE
no mpls ldp sync
!
! Step 3: Optionally enable SR-TE policies now that SR is sole protocol
segment-routing
traffic-eng
! Configure SR-TE policies here
!
!
! Final verification:
show isis segment-routing prefix-sids ! All nodes have SIDs
show mpls forwarding-table ! Only SR labels remain
show isis fast-reroute summary ! TI-LFA coverage (should be 100%)
!
! Confirm no LDP sessions remain:
show mpls ldp neighbor
! Expected: empty (no LDP sessions)
Rollback Procedure
! OcNOS -- Phase 2 rollback (revert to LDP preference)
!
router isis CORE
segment-routing mpls
no segment-routing sr-prefer ! Remove sr-prefer, LDP takes over
!
address-family ipv4 unicast
segment-routing mpls
no fast-reroute per-prefix ti-lfa
exit-address-family
!
! Traffic immediately reverts to LDP labels
! Verify:
show mpls forwarding-table ! LDP labels active again
- Segment Routing Explained: SR-MPLS Fundamentals
- ISIS-SR with TI-LFA: Fast Reroute Guide
- Flex-Algo in OcNOS
- OcNOS-SP Product Page
- Contacter IP Infusion
IP Infusion Engineering Team