Edge Is The New Core

Let’s face it, the debate is over when it comes to whether telcos should create innovative and localized services in the core or at the edge. If past performance is an indicator of future success, it’s clear telcos need a radical transformation of their self-defined legacy network hierarchies. This allows them to get away from rigid network boundaries and focus on net-new opportunities to drive growth and revenues. To create these new opportunities, they must go closer to the edge — where the users are — and have a flexible and nimble architectural approach. While this is a new model from a telco perspective,  it is business as usual for cloud providers, their real competition.

Looking at the 5G vision from the perspective of International Mobile Telecommunications (IMT-2020) requirements, there are three key network focus areas:

  • Throughput – to accommodate eMBB (enhanced Mobile Broadband) use cases
  • Latency – to accommodate uRLLC (ultra-reliable low-latency communication) use cases
  • Massive Connections – to accommodate mMTC (massive machine type communications) carrier IoT

These key areas mandate a different network infrastructure, meaning a shared common infrastructure focusing on the route to market and how the telco chooses to monetize the specific use cases tied to it. The monetization can target B2C use cases or B2B with a vertical focus. The key theme emerging here is that the network infrastructure is designed to drive business outcomes and telcos must look at it through the lens of an end-to-end transformation incorporating the following principles:

  • Architectural adjustmentsEssential Network functions migrating from core to the edge. The idea is to leave the control plane functions in the centralized locations and move the user plane/throughput-intensive functions to the edge. These simple architectural adjustments shave off a good amount of latency and roundtrip traffic time. Having a common cloud fabric that allows the workload positioning on-demand will only increase reliance on containers.
  • Network Infrastructure SharingFocus is on common x86-based compute infrastructure that drives multiple workloads. By decoupling network functions from the processing hardware and a variety of workload software from different vendors, they can share the same compute infrastructure. The workloads can be real-time and bandwidth constrained, such as vRAN or AR/VR, but at the same time able to accommodate a volume of connections such as narrow band IoT (NB IoT) and LTE-M scenarios for industrial IoT use cases. These workloads can coexist on the same shared compute infrastructure and are further segmented, or sliced, to guarantee SLAs and KPIs in a managed services scenario led by the telco.
  • SLA-based services focus Network slicing concept is to create SLA-based service opportunities as network infrastructure becomes a shared resource. While B2B scenarios largely look like a telco managed service, it is an ecosystem or partnership-based service model in the end. Telcos may choose to partner with MSPs or CSPs to create a bundled service offering and therefore the network itself must be flexible to provide this capability. After all, network slicing is a borrowed enterprise concept where it is widely known as multi-tenancy. The idea is very much the same, which is to provide guaranteed services without disruption.

Dell Technologies’ vision is to transform the network through workload virtualization and deliver greater flexibility via software-defined networking and open compute infrastructure with telco-grade servers to meet a variety of deployment options. In addition to these telco-grade, highly reliable servers, platform partnerships with VMware and Red Hat facilitate workload orchestration and provisioning to onboard new use cases quickly. Dell Technologies has a dual-prong strategy related to the 5G Transformation, which is essential for the Telecom Transformation and is integral for the Telecom Digital Transformation.

As workloads demand localized content and traffic flows allow offloading, aggregation and distribution of real-time and cached content, use cases such as vRAN, AR/VR, and AI/ML require a common compute infrastructure with plenty of acceleration methods. This is an area where Dell Technologies is actively collaborating with NVIDIA to bring GPU-based acceleration at the edge.

NVIDIA is building a programmable, cloud-native and scalable edge compute platform stack for telcos. The NVIDIA EGX platform is designed to deliver new AI services to consumers and businesses while supporting the deployment of 5G, virtual RAN infrastructure. This flexibility enables the convergence of B2C/B2B applications and network services in a single commercial off-the-shelf platform.

NVIDIA EGX enables a new GPU class for accelerated AI computing designed to aggregate and analyze continuous streams of data at the network edge. NVIDIA EGX includes an optimized software stack for Dell infrastructure that features NVIDIA drivers, a Kubernetes plug-in, a container runtime, and containerized AI frameworks and applications, including NVIDIA TensorRT, TensorRT Inference Server, DeepStream SDK and others.

Through this collaboration, Dell Technologies and NVIDIA are bringing the most advanced telecom transformation to market and helping our customers to be #5GReadyNow. In fact, Dell Technologies and NVIDIA have launched a Telco focused solution on SwiftStack.

Dell Technologies and NVIDIA are collaborating to transform the telecom market and help our customers to be #5GReadyNow.

Learn more about NVIDIA Aerial and the Aerial SDK Developer Program.