Learning Objectives

• Understand how segment routing simplifies MPLS by removing LDP and RSVP-TE
• Explain the difference between SR-MPLS and traditional label distribution
• Identify when to use SR versus LDP

The Complexity of Traditional MPLS

Traditional MPLS requires multiple control-plane protocols working together. Interior gateway protocols (OSPF or IS-IS) advertise IP reachability. LDP distributes labels for every prefix. RSVP-TE signals explicit paths for traffic engineering. BGP carries VPN routes. Each protocol has its own state, session management, and failure modes.

This complexity increases operational overhead. Mismatched LDP label bindings, RSVP-TE path calculation failures, and protocol interaction bugs are common sources of carrier network outages. Segment routing was designed to eliminate most of this complexity.

How Segment Routing Works

Segment Routing (SR) takes a fundamentally different approach: instead of distributing labels via LDP, the router assigns labels algorithmically based on the IGP shortest-path tree. An SR-enabled router allocates a prefix Segment Identifier (SID) for each IGP prefix — typically from a dedicated SR Global Block (SRGB) like 16000-23999. Every other router in the domain can reach that prefix using that SID, with forwarding paths computed by the IGP.

The key insight: the source (ingress) router controls the path by stacking SIDs. A packet carries a list of segments (labels) that define the exact path through the network. Each intermediate router pops its own segment and forwards based on the next segment in the list. This is called source routing — the path is encoded in the packet, not signaled hop-by-hop.

SR-MPLS in Action

Unlike LDP, where each router independently picks its next hop, SR gives the ingress router full visibility and control. The path can be calculated by a central controller (e.g., a Path Computation Element) or directly on the head-end router. No signaling protocol runs between the transit routers — they just forward based on the top SID using the IGP-derived forwarding table.

SR vs LDP: When to Use Which

Traffic Engineering with SR

Segment Routing's traffic engineering capabilities are one of its strongest advantages. An SR-TE policy defines a segment list that encodes the desired path — including explicit nodes, adjacency SIDs for specific links, and latency or affinity constraints. The policy can steer traffic away from congested links, enforce disjoint paths for diversity, or guarantee latency bounds.

Because no signaling is involved, SR-TE policies can be programmed on the head-end router or pushed from a central controller via PCEP or BGP SR-TE. This makes SR ideal for automated traffic engineering — the network can react to congestion by recomputing paths and pushing new policies in seconds.

Migration Strategy

Carriers rarely rip out LDP overnight. The typical migration involves enabling SR alongside LDP (both distribute reachability to the same prefixes), then gradually shifting traffic from LDP to SR by configuring SR-TE policies or adjusting IGP metrics. The label stack can even mix SR and LDP labels during the transition.

Key Takeaways

• Segment Routing removes LDP and RSVP-TE from the MPLS control plane
• Prefix SIDs are allocated from the SR Global Block and advertised by the IGP
• The ingress router encodes the path as a segment list — source routing eliminates hop-by-hop signaling
• SR-TE enables automated, centralized traffic engineering without protocol overhead
• Migration from LDP to SR is incremental — both can run simultaneously