While promoted as a way to simplify the network stack, the move to a fully cloud-native 5G SA core is not without its challenges. 5G SA introduces new levels of complexity into the network, depriving operators of visibility into the network performance and user experience insights that are so crucial for 5G revenues.
This complexity is the result of it: a) decomposing network functions infrastructure (NFI) into microservices and b) dynamic instantiation and teardown of network functions along the service path as usage ebbs and flows.
Despite these substantial challenges, MNOs are driven to adopt 5G SA's cloud-native core because of the advanced capabilities doing so brings, such as support for virtual (VR) or augmented reality (AR) and mobile gaming, accelerated time to market for new services and private 5G networks.
A Heavy Reading survey found that 49% of MNOs are planning to deploy 5G standalone (SA) within the next year, and a further 39% within one to two years. This is a task described as 'heavy lifting' by BT, a UK service provider.
Services such as AR and VR are highly dependent on quality of service (QoS) and customer quality of experience (QoE). To deliver on these requirements, MNOs are setting latency 'budgets' and then re-evaluating and optimizing each stage of the service path to fit within very aggressive targets.
Private 5G networks come with the same expectations that govern software-defined wide area networks (SD-WAN) in the wired world. Customers of private networks will expect rigorous adherence to contracted classes of service, especially as they take up the latency- and bandwidth-sensitive use cases that network equipment manufacturers and service providers have been trumpeting for years. At Mobile World Live in May 2021, experts forecasted that over 89% of 5G enterprise revenues will be backed by service level agreements.
For example, video game publishers will require aggressive latency thresholds because of the real-time nature of competitive online gaming, where 10-20 milliseconds can mean the difference between scoring a kill or being roadkill. The whole value proposition of providing immersive worlds depends on minimizing the time between actions and outcomes. Developers of AR/VR experiences will demand no more than 20 ms of latency to ensure that the virtual world refreshes in time with physical world movements.
Meeting these latency demands to find profitability with 5G means MNOs must address major changes head-on to service assurance capabilities and toolsets. Operations and network architecture teams are struggling to achieve the same network insight with 5G that they had into their 4G networks. Factors contributing to this trend include the fact that:
a) Thanks to encryption, up to 98% of performance impairments are invisible to traditional service assurance tools, with passive probes in particular struggling to keep pace, according to a Heavy Reading custom survey from July 2019.
b) The volume of network performance and telemetry data has exploded as a result of network densification in urban environments, cloudification of the network stack where compute is the new fault domain and decomposition of the NFI, with inter-process communication handled by the service mesh.
5G services-based architecture (SBA) therefore demands an entirely new approach to troubleshooting and service assurance. It's not just a matter of how data is collected but also how much data is collected. In a cloud-native environment, you need cloud-native service assurance solutions that can be instantiated at all points in the network, up and down the service path. They should also provide the orchestrator with the information required to instantiate new resources to achieve target performance levels.
One leading service provider expects to deploy dozens of probes across its nationwide network, with each network probe expected to capture over 100 Gbit/s of data. This translates into petabytes of data that need to be transported, stored and analyzed. At this scale, it's simply not feasible to analyze all the data in anything approaching near real-time.
The solution is a two-step approach: first, sample data adaptively and second, leverage telecom-specific algorithms and machine learning-enabled analytics to make sense of it by correlating cases and prioritizing resolution based on customer impact. It's this "small data" that helps identify contributors to poor QoE and ultimately customer churn.
MNOs need real-time network, service and quality of experience intelligence that is TCO-optimized. Cloud solutions that don't sample selectively often have a hidden cost: the high levels of throughput and storage that are required to transmit and store all the performance data coming off the network. These incurred costs typically add up to many multiples of the original solution cost, making TCO a real concern for organizations moving to a cloud footprint under 5G SA.
Operators need the ability to monitor per service, per application and per device, while only capturing the data necessary to achieve and restore service levels. Only with this enhanced level of service assurance and support committed to meeting enterprise SLA requirements will MNOs be able to generate the 5G revenues required to fund the network.
Allen Johnston, VP Sales, Americas, EXFO
About the author: Floyd Allen Johnston, VP Sales, Americas at EXFO
Floyd Allen Johnston was appointed as VP of sales for the Americas in September 2019. Floyd has 27 years of experience in the telecommunications industry and extensive expertise in both wireline and wireless technology. Before joining EXFO, he served as vice president of the North America and Americas divisions for Mavenir. He previously held leadership roles at Oracle, Acme Packet and Tekelec. Floyd has an MBA in telecommunications from University of Dallas, and a master's degree in mediation and conflict resolution from Southern Methodist University (SMU) in Dallas, Texas.
Image source: INNA FINKOVA/Alamy Stock Photo