Network segments: Routing to greater success

Segment routing could change the way MPLS networks function and facilitate the adoption of software-defined networking (SDN) while simplifying operator networks in the age of the data deluge.

As a modern approach to MPLS transport, segment routing is gaining popularity as a way to simplify MPLS networks. Network monitoring and automation have evolved to a level where the self-healing network is a viable option for organisations across the world, and downtime is minimised by closing the gap between detection and correction of problems, without human intervention that inevitably takes up time.

At the same time, the modern data centre, cloud and service provider environments are seeing an explosion of traffic and drowning in a myriad of routing protocols, making it necessary to simplify data traffic steering at scale.

One of the primary means of improving the resiliency of a network and solving traffic congestion problems is segment routing (SR), developed by the Internet Engineering Task Force (IETF’s) SPRING Working Group. It is an evolved version of source routing (path addressing) which has allowed a sender of a packet — as opposed to routers — to specify the route the packet takes through the network.

The management of traffic from source to destination in Multiprotocol Label Switching (MPLS) networks has, for a long time, been one of the main issues for reliable and efficient allocation of networking resources. As network topologies become increasingly complex, a simple but intelligent way of channelling traffic has emerged in the form of SR.

Dividing the network into segments — each with its own unique segment IDs consisting of  unsigned integers — helps the source itself do the routing and choose the path of the packet by including an ordered segment list in the packet header. Implementing SR on an existing MPLS network is relatively painless as it does not require additional protocols; another upside is that it provides native IPv6 support across the protocol stack.

Residing in the SR application controller is a function that computes the most efficient path to steer traffic, depending on different requirements and the current state of the network. This path computation function harmonises the tension between centralising and distributing data traffic, the two core characteristics at play in network architectures.

The challenge has always been to avoid complication and maintain flexibility, yet ensure that traffic takes the most efficient path throughout the routes available. SR allows for this without a motley of protocols and rigid pathing inherent in traditional traffic engineering by making it much more granular. It is able to achieve simplicity as well as scalability in unprecedented ways.

SR can also help protect against distributed denial of service (DDoS) attacks — while unsophisticated as a security risk, these have been surprisingly effective at overloading network traffic. When a DDoS attack is detected, SR can collect traffic at a centralised data cleansing station which removes malicious traffic and then channel it back into the network again.

SR can support the overall architecture of SDN and significantly enhance its implementation, and is a definite step towards complete virtualisation. It allows network architects to programme the network, for example, to send voice data over a lower latency path and send bulk data over a higher latency path. This results in a much lower overall latency, as network bandwidth is being used most effectively. Also, if a node or link fails, the inherent resilience imbued by SR helps restore connectivity very quickly.

SR can have massive impact on the delivery of OTT services to customers, and as bandwidth-heavy content becomes more popular — from augmented and virtual reality content to 4K video — network operators will gradually deploy SR in phases over the coming years, singling out use cases and honing implementation skills as it is more fully fleshed out.

Ultimately, SR will allow operators to cater to exponentially rising customer demand and adhere to very strict SLAs while decongesting their networks more effectively. SR may well be the future of MPLS networks in the majority of use cases. It is also a step towards architecting a responsive and self-healing network.