Employing Public Cloud Architectures for Private 5G

Beginning in 2012, operators started expanding data center capacity in support of emerging network functions (NFV) and OSS/BSS virtualization initiatives. With the inefficiencies of simply porting monolithic software onto virtual machines now well recognized, these operators are accelerating the implementation of software virtualization to support network functions (VNFs) built, from scratch, to be purely cloud native. Employing microservices methodologies and instantiated inside containers interconnected by dynamic service meshes, highly automated cloud native network functions (CNFs) can be rapidly deployed and granularly scaled with incredible efficiency.


Even for larger carriers, however, designing, deploying and maintaining these private cloud infrastructures can be a daunting challenge. Leveraging in-house IT skill sets can help, but even experts in these disciplines would not be familiar with the special demands of the compute workloads or the unique architectures required to support real-time communications control and user plane traffic. While such deployments might be achievable in a few large centralized locations, most service provider applications are latency sensitive and include components that must be deployed in a highly distributed manner in relative proximity to consumers or endpoints. Indeed, for a large majority of carrier offerings, there are many quality of service guarantees and service level agreements that must be maintained though high traffic load events and in the event of catastrophic outages. Without careful consideration, this could result in a significant amount of expensive compute capacity and supporting infrastructure which remains unused for a large majority of the time.

Businesses Benefits from Public Clouds

Private cloud infrastructures like Microsoft’s Azure focus on driving down costs and opening new revenue streams by empowering architects to migrate critical service functions to highly distributed compute clouds. Employed for Infrastructure, Platform and Software as a Service (IaaS /PaaS and SaaS) Azure has the most regional locations of any cloud provider. The highly distributed nature of Azure enables Microsoft to deliver on a critical feature: low latency data processing. Recognizing that emerging services are increasingly demanding a degree of ultra-low latency that can only be afforded by highly localized compute, Microsoft also provides the ability to extend Azure to private premises, such as network operator, managed service provider and enterprise locations. Azure Stack is a full rack of servers and switches supporting large on-site deployments of a public cloud, while Azure Stack Edge scales down to a single server.

For communications service providers, the need to adopt a straightforward edge compute strategy has never been so critical. Businesses naturally benefit from leveraging public cloud infrastructures as either standalone services or as part of a hybrid cloud strategy. While the core data center operators built to support IT functions or early NFV projects provide the appropriate economies of scale to host centralized compute functions, public clouds can extend geographic reach either inside or outside an operator’s serving region. These new points of presence (PoPs) form the foundation of a multi-access edge compute (MEC) strategy that enable either network functions or applications, serving the traffic steering and processing needs of regional subscribers or endpoints, to be efficiently deployed and maintained. Furthermore, the on-demand nature of public cloud compute resources and the ability to start small, enables incredibly cost-effective introductory trials, capacity bursting and redundancy.

Enterprise Intersection with 5G

As enterprises rapidly evolve in step with the fourth industrial revolution (Industry 4.0), highly reliable and secure wireless infrastructures supporting thousands of control endpoints and sensors are becoming the expected norm. The applications serving these modern research, manufacturing and distribution organizations require low latency access to dedicated compute resources. Combined, these specific requirements are resulting in increasing interest in private 5G solutions as an enabling technology for such enterprises. 5G enables a far higher density of endpoints than alternatives such as WiFi6, making it ideal for industrial automation and IoT. In outdoor coverage settings, 5G affords the advantage of macro area coverage and high-speed mobility. As a public access technology 5G has better policy controls, cellular-level quality of service and affords greater security, with the option of classical subscriber identification modules (SIMs).

While early private wireless implementations such as the citizens band radio service (CBRS) and LTE-U have employed licensed assisted access (LAA) techniques and required comprehensive listen before talk (LBT) procedures to prevent interference with public Wi-Fi, 3GPP release 16 has taken steps towards private 5G deployments that can be completely standalone (SA). This drive to Unlicensed New Radio (NR-U) is being supported by a slew of unlicensed sub-6Ghz mid-band frequency ranges opening up around the globe. Ultimately, SA NR-U enables systems integrators (SIs) and managed service providers (MSPs) who have no licensed spectrum assets to more readily adopt 5G as the foundation for their enterprise offerings. As such, these outfits have been some of the first to study 5G as an alternative to Wi-Fi in industrial environments.

Release 16 is just the beginning of the 3GPPs focus on industrial applications, with release 17 looking deeper into the vertical markets that could benefit from ultra-reliable low latency communications (URLLC). The 3GPP’s TS22.104 is just one technical specification focused on the specific requirements of the type of physical systems and computational components demanding very low end-to-end latencies and reliable access to control applications. The result of this is a set of communications service reliability and availability (CSR/CSA) metrics around latency, jitter and packet loss, which would form the basis of predefined key performance indicators (KPIs). With the relative dearth of available capacity, the ability to meet these KPIs for wide-scale adoption of NR-U is still dependent on being able to deploy wireless infrastructures that are not subject to crosstalk and interference even in dense deployment areas. This is where the type of LBT technologies used in non-standalone (NSA) private wireless deployments are again being considered. With increases in capacity facilitated by the implementation of spatial sharing techniques, the addition of such localized LBT resource allocation servers support a contention-based system that synchronizes access to the underlying unlicensed access infrastructure across otherwise unassociated endpoints.

Coexisting in Carrier Infrastructures.

Leveraging inherent agility, scale and reach, Azure provides a unique suite of products and solutions targeted to the needs of any network operator, MSP or SI who may be considering 5G NR-U to serve the most challenging applications and demanding enterprises. Azure can be employed as a primary compute platform for hosting network functions or to enable cost-effective redundancy and provide supplemental capacity, purely on-demand. Azure accelerates the adoption of cloud services within network operators by eliminating the need to build out private infrastructure and then dramatically simplifying ongoing operational activities through dynamic orchestration and automation. This model not only disrupts the cost economics typically associated with delivering communications services, it provides a platform for expanding geographic serving areas and delivering new revenue generating services. These include applications requiring ultra-low latency edge compute or complex machine learning algorithms and artificial intelligence. Focused on low latency, high bandwidth, edge compute, the latest addition to Azure - Edge Zones - is specifically targeted to supporting 5G network functions and services in operator and enterprise environments.

The breadth of integrated services Azure offers has the added benefit of ensuring the ability to easily create completely autonomous network slices with distinct service chains comprising independent middlebox functions performing specific actions on user traffic. Beyond basic packet handling, more advanced application layer gateways (ALGs) and virtual private networking (VPN) resources can be combined with distributed mobile network functions to create a complete networking node that is perfect for businesses with sophisticated inter and intra-networking requirements. Correctly implemented, public cloud services provide a superior subscriber experience while reducing the cost of delivering existing services and providing a platform for innovative solutions.