It has been a decade or so since society began to discover the many benefits of the cloud. Since then, online services have changed our lives: how we communicate with others, how we entertain ourselves, how we do business and run the infrastructure that supports us. And yet, in many ways, this is just the beginning.
The cloud — we sometimes need to remind ourselves — is symbiotic with the network. As the cloud has expanded, the network has evolved alongside. Data center technologies which are essential to the cloud, are looking to be adapted for use in the broader underlying network, just as new network technologies are making new kinds of cloud services possible. Also, cloud computing is becoming more dynamic in its capabilities, opening further opportunities for services and applications, which will, in turn, require further network innovation and evolution.
There will be global demand for a 100 times more network capacity over the next decade. This will be driven by our viewing habits moving from linear TV broadcast to ever increasing personalized unicast consumption along with augmented and virtual reality adoption (AR/VR). In the future, latency and service level agreements (SLAs) will matter as much as bandwidth.
Cloud applications are also moving into new territory as organizations embrace them for manufacturing and industrial applications (4IR), smart cities and machine-to-machine communications. Large-scale deployment of connected sensors and actuators will enable the deployment of mission-critical control systems such as the automation of manufacturing and industrial processes, large-scale coordination of autonomous vehicles, and logistics and smart distribution grids.
This broad range of new applications and services will also add to the capacity demand, driving the network to deliver stricter service-level agreements (SLAs) and dynamic agility, flexibility and responsiveness – as well as drive a more distributed architecture. Lower latency requirements, especially in the areas of industrial automation, machine communications and interactive multimedia, such as AR/VR, will demand that certain network functions be placed closer to the application site or end user. Neither centralized data centers nor cores can meet the need for latency delays of less than 10ms. Much like content delivery networks, these applications will require edge clouds that handle some of the tasks locally.
This new distributed cloud architecture will have to be mirrored in the architecture and design of the network. From a network core perspective, a universal, adaptive and multi-access cloud-native core will be needed to ensure the delivery of the varying service and application requirements.
To support this, an architectural evolution has been occurring in mobile networks as we’ve progressed through 2G, 3G, and 4G. With the later releases of 4G, which anticipate 5G, we have already seen the advantages of continued architectural evolution with the further separating of the control and user, or forwarding, planes. This has been standardized in 4G with CUPS (control and user plane separation) and is inherent in the new 5G 3GPP Release 15 core specifications.
What this separation makes possible is the expanded flexibility to distribute specific core network functions where they are needed. For instance, autonomous vehicles will demand elevated levels of reliability and extremely low latency from a connectivity network. Thus the need for distributed and high-performance user planes situated closer to the edge of the network.
A further new core innovation is the ability to ‘slice’ core network resources to provide customized networks and deliver differentiated treatment depending on user requirements, which helps to improve the efficiency of network resources. This capability is available in 4G network cores with DÉCOR/eDECOR, and will be extended for a more end-to-end (E2E) network operation in the forthcoming 3GPP Release 15 5G core.
This will allow core resources to not only be distributed but also to be allocated to specific services/application types (vertically) or specific customers (horizontally). Vertical functions might include voice, content distribution, web services, autonomous car support or other IoT device types. Horizontal slicing might allocate core resources to specific enterprise customers, allowing them to effectively extend their private enterprise networks across the applicable access technologies.
To monetize network slices as services, slices will need to become elastic to deliver varying service requirements. This will require network intelligence to capture the E2E state of the network and trigger real-time proactive actions, including resource orchestration, self-configuration, healing, and optimization.
The disaggregation and distribution of network functions might not entail a purely virtual approach. Just as we are seeing the increased use of task-specific processors in the data center, some functions, such as data forwarding, might best be handled by hardware-based network processors. Given the bandwidth and latency demands of some services and applications, dedicated network-processor-based physical user plane functions may deliver better cost performance benefits, and help reduce footprint and power considerations in data centers or other locations.
It would also be extremely beneficial if these user plane functions could deliver a rich set of built-in IP networking and advanced services capabilities, such as high-density VPN and tunnel termination, network gateway user plane, firewall and carrier-grade NAT functions. This would help remove the requirement for the deployment, integration, and management of additional network functions.
To summarize, as the cloud evolves, so too does the network — and vice versa. Much of the cloud’s early development exploited the pre-existing network. The initial networking evolutions needed by the cloud were restricted to within the data center itself. But as the cloud matures, we need to re-architect the underlying network too.
As we move to evolve our networks, the changes to the network core are going to be profound. The new core design will need to embrace cloud technologies and cloud-native architectures with continued software disaggregation, with the flexibility to support ‘core slicing’ and a more distributed network function deployment. It will also need to be capable of mixing the appropriate virtual or physical network functions to meet the specific operational and cost requirements of the broad range of services and applications, along with adjusting in real-time to shifting demand patterns.
As an industry, we will need to think about how best to evolve the cloud and the network symbiotically, if we are to truly support and enable innovative cloud-based applications and services that will shape our lives in many new and exciting ways.
With over 20 years of experience in the telecommunications industry, Marketing Director Nick Cadwgan has held senior architectural, product marketing & management, and business strategy roles focusing on Broadband Access, Carrier Ethernet, Carrier/IP Routing, Optical Transport and Mobile Networks with Motorola, Nortel Networks, Newbridge Networks and other privately funded companies. Nick brings a proven combination of marketing, technology and business management expertise to his current role at Nokia.
