5G was conceived to improve multiple facets of network performance – higher bandwidth for sure, but also lower latency and high resiliency. In effect, 5G brings to the consumer domain service levels that were so far reserved for business. The Covid-19 pandemic, which is keeping many people at home, is accelerating this trend. Consumers are more willing than ever to pay a premium for differentiated services and stability. Frequent hiccups, video call interruptions, and unreliable bandwidth are no longer tolerated by subscribers at home.
5G technology solves many of these issues by changing mobile networking principles:
- Instead of steering all user traffic indiscriminately through the same gateways, the physical network is sliced and treats traffic flows according to individual application needs.
- Application content (streaming, gaming, IoT) can be located at the network edge instead of centrally. The user plane (i.e., content) is split from control plane (user ID, access control) and flows from edge data centers that dynamically place content closer to mobile devices.
- Radio spectrum is increased dramatically by using millimeter wave bands.
Since 5G architecture is not hauling everything back to the core, 5G networks are more distributed and more dynamic, constantly adjusting as they go. To meet the strict demands of different applications and services, 5G networks must fully leverage the benefits of virtualization and maximize the utilization of network assets. The workload of content servers and mobile core components need to be widely distributed and able to utilize resources dynamically on demand. New 5G base stations (gNb) have to be connected to distributed mobile core components which requires multiple connections over multiple 5G network slices – each with a different set of configurations.
In short, 5G is complicated. And it calls for much higher levels of network automation. The ability to automate relies on certain capabilities, including up-to-the-second reliable view of network resources, intelligent algorithms for service path computation based on measured performance and available resources, and smart network control methods.
While some 5G networks will be built from scratch to achieve these goals, most carriers will gradually shift from 3G-4G to the new technologies using add-on enablers. One such enabler is Segment Routing, which is a flexible, scalable way of doing source routing and has been championed by Cisco for last couple of years. In a recently published blog, Cisco explained the ability of segment routing to lower latency because it uses existing IGP/BGP routing protocol with new metrics while allowing the source router or a centralized PCE to determine the service path from A to Z. The source understands its needs and therefore sets the explicit path through all routers between A and Z.
To provision a preferred 5G path, new metrics are added to IP links in addition to IGP and TE. They are set with measured delay and Shared Risk Link Group (SRLG) and allow the path computation algorithm to optimize by any metric and to use affinities as link or node IDs.
The concept of Segment Routing provides a very good way to “bridge” between older routing methods and 5G. It’s a much needed enabler for carriers who are migrating. However, Segment Routing is missing a crucial element for successful implementation – accurate and complete network inventory and connectivity data. Without accurate data, the new metrics and computation algorithms to guarantee low latency, bandwidth and diversity will not be able to meet the stringent service path requirements of 5G network slices. For example:
- Latency – The segment-routing latency metric is based on a measurement in the IP layer. That helps when desired latency can be achieved but when such is not achievable, then knowing the optical path is a must. latency can creep in and accrue over long distances and the way to solve it is to change IP link path to go over shorter optical paths, which is out of segment routing scope.
- Resiliency – The SRLG metric is supposed to avoid shared risks and to guarantee diversity between primary and secondary path. However, most SRLG data are based on manual update by technicians who work with a spreadsheet received from engineering or with a manually updated fiber management tool. If the data are not correct, neither will the SRLG metric be.
- Bandwidth – To make sure that desired bandwidth can be provided, and no bottlenecks are expected, the bandwidth as a resource needs to be managed. Based on measured traffic utilization and provisioned bandwidth, one can decide if selected path has the sufficient bandwidth. NetFusion manages bandwidth and enables considering it when selecting a path for new service. Traffic utilization as a metric based on a pattern calculated on data collected over time is more accurate that just simply taking a snapshot of the current traffic level.
Sedona NetFusion closes these gaps and complements segment routing. Sedona NetFusion Network Intelligence and Automation platform uniquely discovers inventory in all network layers, from fiber to L3 services, including the stitching points between optical and IP networks. Using accurate, real-time network data, Sedona NetFusion is able to calculate latency and diverse paths across all layers of the network, assuring that SLA demands will be met with zero dependencies on partial information or manual inputs.
Sedona NetFusion Controller unlocks the benefits of segment routing by computing paths centrally and Propagating them to the network as segment routing policies.
- With the combination of segment routing and Sedona NetFusion, network configurations remain simple and easy to manage while maintaining optimal paths based on centralized computation.
- With Sedona NetFusion, there is a 100% guarantee that stringent service SLA requirements will be met all along the multi-layer service path, ensuring low latency and strict fiber-layer diversity for high resiliency.