During September, I’ve been very fortunate to attend two actual physical conferences focused on 5G. The first was in Denver Colorado, conveniently just across the Rocky Mountains from my current home in Grand Junction. The second was in London England, 200 miles or so south of my original home in Yorkshire (typically celebrated as the home of the most successful cricket team ever, though puddings also get a well-deserved mention).
Where’s the Return on Investment for 5G?
As Communications Service Providers (CSPs) worldwide scale up the deployments of their 5G networks, they face strong pressure to optimize their Return on Investment (RoI), given the massive expenses they already incurred to acquire spectrum as well as the ongoing costs of infrastructure rollouts.
Like previous “G”s, 5G is filled with promise and with the potential to enable a vast set of new applications and services that are envisioned for both enterprises and consumers. However, also like with the generations that preceded it, CSPs are saddled with the immense burden of up-front costs for new spectrum and the supporting infrastructure. While CSPs will reap the benefits of 5G for a decade or more, they must also focus on the cost of the network infrastructure as it scales in users and bandwidth until those profitable and newly-imagined services emerge and grow.
Server utilization matters
5G networks have a new architecture compared to 4G. The 5G packet core is implemented as a set of Virtual Network Functions (VNFs) or Cloud-native Network Functions (CNFs) running on standard servers, rather than using purpose-built appliances. The cost of these servers, both CAPEX and OPEX, is a major contributor to the overall cost of deploying and operating 5G networks, so minimizing server costs is key to improving RoI.
As the primary data plane function within the 5G packet core, the User Plane Function (UPF) represents the highest compute workload. Performing critical functions associated with connecting user and device traffic from the Radio Access Network (RAN) to the Data Network (DN), the UPF is responsible for packet inspection, packet routing, packet forwarding and Quality of Service (QoS) handling.
Since general-purpose server CPUs are not well suited to the performance and latency requirements of real-time packet processing, CSPs and 5G core software vendors typically adopt solutions for offloading the UPF to accelerators such as programmable Smart Network Interface Cards (SmartNICs) which are optimized for executing such workloads. These SmartNICs replace the standard NICs (often known as “foundational NICs”) used in IT-oriented data center servers, providing orders of magnitude better UPF performance than achieved using the server CPUs themselves.
SmartNIC offload technology maximizes packet core RoI
Leveraging our team’s expertise in delivering performance-optimized software packages integrated with leading-edge SmartNICs, Napatech has launched a UPF offload solution that provides best-in-class UPF performance. The solution comprises a fully-offloaded UPF fast path implemented within the Link-Inline™ software stack, running on SmartNICs that support a bandwidth of either 100Gbps or 200Gbps, depending on the card selected. Using a single 200Gbps SmartNIC to sustain 100Gbps of full duplex traffic, this solution processes up to 100 million concurrent flows, with a flow learning rate greater than 1.5 million flows per second, achieving a total throughput of up to 85 million packets per second and ensuring full wire speed operation for typical packet sizes.
So what do these impressive-sounding performance numbers mean in terms of actual meaningful cost metrics? It turns out the answer is extremely interesting if you’re a CSP looking to optimize the RoI for your packet core data centers. An analysis of a metro edge data center running a 5G packet core to support 50,000 5G users, with OPEX calculated over a five-year period and including the cost of power as well as server OPEX, yielded compelling results: the Napatech UPF offload solution enables CSPs to support 7x more users per server than an ASIC-based NIC, with an 82% reduction in per-user CAPEX and an 86% reduction in per-user OPEX.
Of course, your mileage may vary, so our team at Napatech is ready to sit down with you and help you analyze the benefits for your specific use case and assumptions.
It takes an ecosystem
Within the telecom industry where ARPU has been steadily declining for the past decade, these levels of capacity increases and cost savings represent an attractive value proposition for CSPs. But CSPs themselves typically don’t purchase individual components like SmartNICs, preferring to rely on Telecom Equipment Manufacturers (TEMs) or System Integrators (SIs) to deliver integrated solutions. So we’re working with software vendors as well as server suppliers, TEMs and SIs, to ensure that our hardware and software are available as part of end-to-end, pre-validated solutions ready for deployment.
As 5G coverage becomes more widespread around the world, technology such as Napatech’s UPF offload solution becomes increasingly important in enabling CSPs to keep their costs under control while waiting for that long-anticipated improvement in ARPU. We can expect more announcements in this space as evaluations and trials progress to partnerships and deployments.