In a short time, we have gone from a world that was once focused on mobile consumers using analog, two-way phone communications to connecting a software-driven, multilayer fabric of devices that will drive innovations in health care, manufacturing, transportation, retail, entertainment, finance, and agriculture. With 5G, the focus is no longer connecting just mobile phones but billions of devices and applications. Service providers face the challenge of updating fixed and wireless infrastructure, data centers, and base stations to build out true 5G while maintaining legacy architecture at a significantly reduced total cost of ownership (TCO).
The True Cost of 5G
According to the Linux Foundation State of the Industry Report in 2021, tremendous infrastructure investments are needed to support the growing device and infrastructure edge demand. The report estimates that between 2019 and 2028, cumulative capital expenditures of up to $800 billion will be spent on new and replacement IT server equipment and edge computing facilities.
Yet, the reality is that while smartphone users generate upwards of $50 on average a month, connected devices may result in only $2 to $3 of revenue per month, per device. To realistically support global bandwidth demand, we need technology that costs an order of magnitude less per connected device and per gigabit of traffic, or we will not bring 5G applications to market in a business viable fashion much less support the growing demand for traffic using 3G/4G networks.
Service Availability Is Critical and Costly
In the era of 5G, numerous applications are emerging, including virtual and augmented reality (VR/AR), industrial robotics/controls as part of the IIoT, interactive gaming, autonomous driving, and remote medicine, to name a few. These applications require a latency, cost point, service availability, and capability to work at scale that simply cannot be delivered via the use of virtual machines running in a typical public cloud infrastructure. Many of these new applications require an end-to-end latency below 10 milliseconds; however, typical public clouds are unable to fulfill such requirements. A modern, more adapted, cloud-based IT service infrastructure is required to address the real-time, high-bandwidth, latency-critical applications distributed at the edge of the network. Service availability became even more important during the COVID-19 pandemic when networks were put through a stress test.
With respect to the core network, service availability is driven at a macro-level to ensure network availability to the consumer of the service through careful planning, software solutions, and geographic redundancy. If a base station crashes for whatever reason, it must be able to request to connect through another base station. In the milliseconds it takes for this to happen, it must search for and reconnect to another service center. If there are several failures, it must expand to another data center in a typical availability zone — this is known as geographical redundancy. Placing physical servers in geographically diverse data centers to safeguard against catastrophic events and natural disasters while load balancing traffic for optimal performance is costly. If you take this model and push it to the edge of a network, guaranteeing reliable service becomes infinitely more challenging, especially considering the billions of new devices and computing resources that will become interconnected. Consequently, compute power must be brought closer to the edge of the network.
Edge Computing and Containers to the Rescue
Distributed edge computing means spreading the network among multiple devices to allow data processing and service delivery to happen closer to the data source or computing device. This proximity to the end user, whether a consumer using a cellphone, a retailer using a point-of-sale system, or an autonomous tractor in the field, is necessary to provide time-sensitive and efficient operations.
Edge computing allows organizations to decentralize services by moving key components of their application to the network's edge. By shifting intelligence closer, organizations can achieve lower network costs and better response times. It is a macro-shift from “silos” to using the same orchestration layer for network, compute, and storage so that management of resources at the edge is simplified and optimized.
For example, one problem is that much of today’s legacy networks run on servers that rely on large virtual machines (VMs). A VM is software that runs programs or applications without being tied to a physical machine. Because latency tolerance or delay for 5G applications is low, VMs are not ideal for handling the demands of tomorrow's network alone. If there is an unexpected outage, it can take several minutes to restart VMs. For this reason, containers are critical. Containers take VMs a step further and are smaller, single, virtualized applications that function as isolated, lightweight silos to run applications across diverse environments on-premises and in the cloud. They make it easier to develop, deploy, and manage apps. More importantly, they can start in milliseconds versus minutes, making them ideal for apps that need to adapt to changing demand.
A Slice Is a Fully Isolated Domain Identity
Demand is dynamic. Standalone 5G networks will need to serve customers with very different needs from those of 3G/4G networks. For example, while applications, like smart-parking meters, value high reliability and security, they don’t require minimal latency. Other applications, like autonomous driving or IIoT, require ultralow latency and high data speeds.
5G network slicing takes the physical network infrastructure and slices it into multiple and fully isolated entities called a slice, vFabric, or virtual distributed data center. By subdividing a physical network into a virtual distributed data center, it creates a fully isolated domain identity. And within this domain, you can create as many subnetworks as you need.
Today, network virtualization (VLAN or VxLAN) can create networks that sit on top of the physical network, these support the virtualization of a network in the cloud while addressing the needs of multitenant data centers by providing the necessary segmentation on a large scale. Multitenant data centers, or colocation data centers, help to reduce costs by allowing organizations to rent space to host their data.
Take, for example, a VLAN that can support 4,096 subnetworks — enough for small to medium use cases. But, this can quickly become a limitation. Theoretically, with overlay networks, such as VxLAN or GENEVE, one can create more than 16 million different subnetworks. This may sound like a lot, but it is still not enough for 5G, and they use tunneling protocols to send data, which is effective but less secure.
Physical 5G network slicing is different — it is an order of magnitude greater. Slicing can partition a physical data center into multiple virtual distributed data centers, and each of these slices can support its own VxLAN. Thus, one data center can now function as two fully independent data centers in a cloud native environment. Within virtual data centers A and B, service providers can split into millions of virtual subnetworks optimized for a specific business case, such as network sharing between operators, supporting an M2M core, or a large enterprise customer. Data transfer is fully isolated, meaning only a unique slice will ever receive data packets assigned to it.
When done correctly, even competitor mobile network operators and data centers can share the same physical infrastructure, divide it into fully isolated slices, and subdivide these into millions of highly scalable virtual subnetworks based on customer service level agreement (SLA) for varying data speed, quality, latency, reliability, security, and services. All run on a fully programmable software-defined edge.
A Multivendor 5G Ecosystem
Today’s providers simultaneously face challenges of limited space and infrastructure and increasing demands on power and cooling. One of the strengths underlined by 5G network slicing is the ability to provide a truly multivendor environment. With slicing, service providers can partition the physical network while offering complete isolation and security in a more efficient and economical way. Breaking down the cost barrier opens the door for greater innovation, new use cases, and business opportunities, creating a more open 5G ecosystem.