91桃色

Networking for AI will be driven by simplicity, speed, and scale. While legacy transport technologies such as MPLS-TP, RSVP-TE, and LDP have served well in the past, it is Segment Routing (SR) that is key to building scalable transport networks for AI and cloud. Ciena鈥檚 Don Jacob discusses how network operators can seamlessly migrate from legacy transport technologies to SR without rip-and-replace of their transport infrastructure.

A network operator’s journey to a Segment Routing (SR)-enabled network is often unique and depends on multiple factors including existing implementations and physical network architecture topology.  One of the key considerations for network operators when planning their network architecture evolution is how to introduce Segment Routing into their networks without significantly disrupting the existing transport infrastructure.

Before we explore that, let’s talk about why you would choose Segment Routing in the first place.

Why Segment Routing?

Evolving networks demand increasingly higher thresholds of scalability, programmability , and simplicity.  Traditional transport technologies like LDP, MPLS-TP, or RSVP-TE are not capable of meeting these thresholds, let alone reducing operational costs.

SR addresses these challenges with its simplified control plane and ability to scale efficiently. Its ability to summarize routes, reduce advertisements, and improve efficiency , along with support for Flex-Algo, SR-TE, and network programmability, makes SR the ideal choice for building future-proof networks.

SR comes in two flavors: Segment Routing MPLS (SR-MPLS) and Segment Routing over IPv6 (SRv6). While both flavors offer significant benefits, , SRv6 is increasingly seen as the end-state for large networks as networks continue to scale. . Its native IPv6 support enables greater scalability, and the removal of BGP Labeled Unicast (BGP-LU) further simplifies the control plane compared to SR-MPLS.

The path to SR transition – interwork, coexist, or migrate?

The path for each network operator to get from a legacy transport technology to introducing SR in their existing network varies. Operators can either choose to let their legacy technologies coexist with SR before a complete migration or where coexistence is not feasible, opt for interworking.

The following table summarizes typical transition strategies and migration paths based on Ciena’s field experience.

Table 1: Migration path for legacy transport to SR-networks based on Ciena's field experience

Ciena’s experience with multiple customers has shown that coexistence is the most optimal approach in terms of reducing both operational complexities and costs. It allows legacy transport technologies to coexist along with SR, ensuring service continuity and protecting CAPEX. With coexistence, operators can continue using their existing hardware without the need for a rip-and-replace, provided their existing hardware supports both legacy transport technology and SR.

Migrating from SR-MPLS to SRv6

As SRv6 capabilities mature and SRv6-capable hardware becomes more common, early adopters of SR-MPLS are considering how to migrate to SRv6. SRv6 delivers flexibility and scalability, derived from its network programming capabilities and massive addressing space enabled by native IPv6 integration. This makes it ideal for transport networks that deliver AI workloads requiring bandwidth-intensive, low-latency services.

Coexistence between SR-MPLS and SRv6 can be achieved through interworking gateways, SR policy stitching for SR-MPLS to SRv6 mapping, and IPv4/IPv6 dual-stack operations — allowing operators to migrate incrementally while protecting their MPLS infrastructure investments. In IPv6-ready or greenfield deployments, SRv6 delivers simpler end-to-end provisioning and automation, enabling the transition towards an intent-driven, SLA-based transport architecture.

The next evolution in SR

When talking about the next evolution around SR, operators are increasingly interested in BGP Color-Aware Routing (BGP-CAR) and BGP Classful Transport (BGP-CT) for its intent-aware, multi-domain path selection that enable efficient network resource utilization and cost optimization. While standards for these technologies continue to evolve, they align with modern network technology such as Source Packet Routing in Networking (SPRING), EVPN, Flex-Algo, and SRv6, enabling end-to-end automation and SLA compliance across domains— key to delivering distributed AI services and applications.

Learn more about transitioning to Segment Routing

Ciena's Jan Straznicky, Senior Director of Product Line Management at Ciena, covered this in his presentation, The Path to Scalable Transport with Segment Routing, at MPLS & SRv6 AI Net World Congress 2025 in Paris, France. Jan talked about the different transition paths, the challenges, and how network operators can migrate seamlessly to SR. Check out the video to learn more.

How Ciena helps you transition to SR

As more networks prepare to implement SR, Ciena can be a key partner supporting the network operators transition to SR-enabled networks with confidence. Ciena has a portfolio of routers of various form factors and capacities that support Segment Routing. Our SAOS network operating system delivers a unified operation experience from the core, across the transport, to the access, and our Navigator Network Control Suite (NCS) eases end-to-end management complexities. Finally, Ciena’s ground-level experience building SR networks for leading network operators has given us the expertise and skills required to help you in this journey.

Segment Routing is the future  of network transport, due to its simplicity and efficiency. SRv6 brings further advantage over SR-MPLS with its scale and programmability aspect. Be it SR-MPLS or SRv6, greenfield or brownfield networks, through coexistence or migration, the path forward is clear: adopt Segment Routing to build resilient, future-ready networks.