Multiprotocol Label Switching (MPLS) for years has served as the technology of choice for network operators to ensure reliable connections for real-time applications, such as Netflix, Webex or Microsoft 365.

You may have heard about the imminent demise of MPLS at the hands of SD-WAN — a more cost-effective approach to networking. This is only partially true.

Yes, MPLS will become obsolete, but not because of SD-WAN.

SD-WAN brings a lot of cost and OPEX benefits to the table but in many cases still utilizes the public internet along with MPLS services for site-to-site connectivity to better ensure delivery, consistent traffic throughput, and latency through the use of Quality of Service (QoS) and fast reroute technologies.

When implementing services with QoS over the public internet, a symmetric internet connection that does not traverse multiple hops via multiple ISPs is recommended. Even if this requirement is met, the delivery over the internet is never guaranteed, and ISP outages or bandwidth bottlenecks can result in excessive jitter and delay and can also force the ISP to discard the best-effort internet traffic. In addition, QoS when implemented over the public Internet is actually edge queuing since there are no end-to-end QoS mappings, like MPLS, implemented across the WAN by the ISP to guarantee high-priority traffic delivery.

Although the public internet has become much more robust in recent years, a customer cannot always guarantee traffic will not be delayed or even dropped as there is no Service Level Agreement (SLA).

Traffic delays and drops are often deal breakers for real-time applications — particularly those utilizing new 5G-standard Ultra Realtime Low Latency Connections (uRLLC), which have a delay requirement of between 1 and 25 milliseconds. As 5G becomes more widely adopted, such latency standards will only become more sought after.

For these reasons, MPLS services will continue to exist to provide guaranteed delivery with real-time fast reroute protection.

Fast Reroute Protection resulting from a failed link.

Large service providers, enterprises and financial and federal institutions will for the foreseeable future need high-speed backbone networks that require high throughput with guaranteed delivery and fast convergence for mission critical real-time systems.

But the challenge remains: MPLS is complex and SD-WAN may not be an alternative solution for every organization.

Enter: Segment Routing.

The next generation of MPLS

Segment Routing accomplishes the same thing as MPLS but is less complex. Better yet, it leverages existing MPLS services and hardware, meaning it does not require new infrastructure and provides an easy migration path.

Segment Routing, which was first introduced by Cisco in 2013, makes the network more scalable and intelligent while improving capacity utilization, leading to lower cost and greater user satisfaction.

The architecture of Segment Routing is based on the source-routing paradigm. It leverages source routing by providing a simple, stateless mechanism to program the path a packet takes through the network.

Podcast: Avoid Traffic Congestion With Segment Routing

Because the application has complete control over the forwarding path and steers the packet through the network by encoding an ordered list of segments in the packet header, there is no need for path signaling. Therefore, Segment Routing does not create any per-flow state and can scale infinitely without any limitations.

Stronger, simpler and more agile

Segment Routing is extremely robust and can deliver sub-50-millisecond fast re-route for link, node and Shared Risk Link Groups (SRLG). Traffic is automatically rerouted to the post convergence path and avoids any intermediate state or utilization of arbitrary links in the network. The primary backup paths are computed automatically by the Interior Gateway Protocol (IGP) with no advanced planning required.

The first iteration of Segment Routing, referred to as SR-MPLS because it relies on the MPLS data plane for forwarding, is backward compatible with MPLS and can be integrated seamlessly into a MPLS environment. It does not require any additional protocols and, in fact, eliminates protocols and simplifies the network.

The second iteration of Segment Routing, called SRv6, further simplifies the network by eliminating MPLS altogether by relying on the native IPv6 header and header extension to provide the same services and flexibility as SR-MPLS, directly over the IPv6 data plane.

Traffic engineering

Segment Routing Traffic Engineering (SR-TE), meanwhile, offers a whole new level of simplicity, scalability and flexibility by incorporating a combination of centralized control and optimization with distributed intelligence.

Traffic engineering in the past was very complex and required a soft protocol like Resource Reservation Protocol (RSVP) to constantly update and calculate the end-to-end path, making MPLS-TE very painful and slow.

Node A choses to direct traffic over a High bandwidth path or low latency path.

SR-TE can be used to steer traffic along any desired path in the network, thus permitting the network operator to utilize disjoint paths or constrain traffic to a low latency forwarding path.

SRv6 can also accomplish traffic engineering directly over a native IPv6 backbone or even the IPv6 Internet, which will simplify network infrastructures by allowing traffic to be steered end-to-end over a predetermined path by the application.

We can program the path directly into the packet header at the source, either via MPLS labels with SR-MPLS, or directly into the IPv6 header with SRv6. The path also can be programmed across autonomous routing domains that have no inter-domain awareness or even IP reachability by incorporating a centralized controller made aware of the end-to-end path via topology information feeds from multiple domains.

The policies required for traffic engineering can be automatically derived and instantiated on demand at the headend and can be either destination- or flow-based, which allows for quality of service to be used as a differentiator when determining the end-to-end path.

Furthermore, there is support for end-to-end real-time delay measurement so that the application can choose the path with the minimum cost, the minimum delay or the minimum cost with a bound delay.

The way of the future

Segment Routing is here to stay and is the future of MPLS.

The advent of SRv6 will accelerate the adoption and migration of IPv6 allowing for greater control and flexibility than ever before. Modern networks demand application control and Segment Routing can provide operators that opportunity.

Connect with me if you'd like to discuss MPLS or segment routing as they relate to your organization.

Technologies