With almost 60 percent of the world’s population now online, traffic continues to grow.
Not long ago, it was a foregone conclusion that mobile devices would cannibalize all the Internet traffic from wireline networks. While slightly more than half of data traffic is generated via mobile devices, the industry accepts that both wireless and wireline networks should work hand-in-glove.
Network operators are considering how to converge wireline and wireless operations. These concepts, called fixed-mobile convergence or wireless wireline convergence (WWC), are based on 3rd Generation Partnership Project (3GPP) standards as well as Broadband Forum (BBF) standards and architectural definitions. Intel has a long history of open innovation in cable, wireline and 5G wireless core networks, which gives us a unique perspective on convergence, among other topics.
Today, data traffic throughput and availability are still the most important key performance indicators (KPIs) in any network, but operators are increasingly focused on improving power efficiency to reduce costs and support sustainability targets.
With Network X just having completed, it’s the perfect time to recap Intel’s work to drive advances in throughput performance, run-time power management, and WWC based on proposed BBF and 3GPP 5G standards. I will share where we’ve been, so I can preview where we can go together.
Breakthrough Performance for BNG, AGF Workloads
Intel continues to deliver breakthrough performance and power with 4th Gen Intel® Xeon® Scalable processors designed and optimized for a range of broadly deployed network workloads and environments, delivering lower latency, higher throughput, deterministic performance, and extended supply life. These processors have more cores, more cache, more memory bandwidth, more + faster IO, and more accelerators and instructions in addition to the architectural advantages from a balance of compute, IO, and memory.
In recent months, we worked with ecosystem partners to demonstrate what’s possible with two different network-SKUs from the 4th Gen Intel Scalable family:
- Intel Platinum 8470N (52 cores with 300 W TDP, Base Frequency of 1.70 GHz, Max Turbo Frequency of 3.60 GHz, and 97.5 MB cache)
- Intel Gold 6438N (32 cores with 205 W TDP, Base Frequency of 2.00 GHz, Max Turbo Frequency of 3.60 GHz, and 60 MB cache)
Casa Systems, Dell, and netElastic each tested virtual Broadband Network Gateway (vBNG) workloads on servers using the Intel Platinum 8470N and Intel E810 Ethernet Network Adapters.
- Casa Systems and Intel testing achieved 1 Terabit per second (Tbps)1 of throughput using the Axyom vBNG on a single server, powered by Intel Platinum 8470N processors.
- netElastic and Intel also demonstrated 1 Tbps2 throughput with vBNG workload in a 2 rack unit (RU), two-socket server powered by Intel Platinum 8470N processors and Intel® E810 Ethernet Network Adapters.
- Earlier this year, tests with Dell demonstrated up to 931 Gigabit per second (Gbps)3 throughput for the vAGF workload using Dell PowerEdge R760 server running on Red Hat Enterprise Linux and Intel Platinum 8470N and Intel E810 Ethernet Network Adapters.
Different test configurations, such as packet size and others, contributed to variations in test results.
Enhancing Power + Performance with Cloud Native BNG Architecture
Most network operators have deployed mid-stack Intel SKUs over the years because of price/performance, datacenter thermals, resiliency, and redundancy leadership. This remains the case today. Intel tested throughput and power performance using Intel Gold 6438N processors, Intel E810 Ethernet Network Adapters, and Intel Infrastructure Power Manager software.
The Intel white paper published earlier this year details its lab tests that achieved 710 Gbps4 in throughput for BNG workloads using Intel Gold 6438N processors and Intel’s vBNG application reference architecture.
The test also delivered up to 14 percent server power savings4 over a typical 24-hour traffic period, with peak server power savings of up to 25 percent4, using the advanced power management capabilities of the 4th Gen Intel Scalable processors and Intel Infrastructure Power Manager.
Showcasing Industry-Leading Innovation at Network X
Intel has highlighted technologies and solutions that demonstrate the art of possible at the Broadband World Forum/Network X events over the years.
In 2021 and 2022, we focused on implementation of subscriber session steering based on the Broadband Forum Working Text 474 (WT474) and cloud native, most recently participating in a demonstration with BDISN and Vodafone to showcase viability, added features and use cases. These early ecosystem engagements are intended to prepare for final ratification of the standard and provide feedback to the open source community for further refinement and development. Vodafone and Intel published a whitepaper on the viability of cloud native and service-based section with BNG/AGF workloads.
Intel has also demonstrated the possibilities with wireless wireline convergence (WWC), which will naturally intersect with cloud native network functions (CNFs) and architectures required for advanced 5G services through standalone (SA) 5G networks.
This 2021 whitepaper outlined the cloud-focused AGF deployment using the Container Bare Metal Architecture (BMRA) to optimize data plane performance and enabling end-to-end data connectivity with the 5G core network. The paper highlighted gen-over-gen performance with 2nd and 3rd Gen Intel Xeon Scalable processors and Intel Network Ethernet Adaptors.
Intel, working with Deutsche Telekom and Mavenir, showed the latest advances in WWC, with wireline Access Gateway Function (AGF) connecting to 5G Core User Plane Function (UPF) in a functional demo at Network X 2023 based on the latest 4th Gen Intel Xeon Scalable processors, connectivity, reference architectures and more.
We will continue to lead efforts across open ecosystems, equipping the industry to turn the art of possible into real-world solutions and deployments.
Andrew Duignan is a Platform Solutions Architect at Intel, working on wireline broadband and wireline wireless convergence. He has previously worked in platform applications engineering role focussed on high speed packet processing on Intel Xeon server platforms. Prior to this he worked as a network software engineer at Intel, and also held roles in telecoms software development and customer technical support in Motorola in both Ireland and Israel. He holds a Bachelor of Engineering degree in Electronic Engineering from University College Dublin, Ireland.
Notices & Disclaimers
Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex
Performance results are based on testing as of dates shown in configurations and may not reflect all publicly available updates. See backup for configuration details. No product or component can be absolutely secure.
Your costs and results may vary. Intel technologies may require enabled hardware, software or service activation. © Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries. Other names and brands may be claimed as the property of others.
2 Tests conducted by Intel on Dec. 7 & 8, 2022. Configurations are available in the addendum section of the whitepaper titled, “netElastic vBNG Reaches 1 Tbps Throughput in Tests with Intel”
3 Tests conducted by Intel on 05/07/2023. Configurations details are available in addendum section of the whitepaper titled, “Wireline and Wireless Convergence Solution on 4th Gen Intel® Xeon® Scalable Processors and Dell PowerEdge R760”
4 Tests conducted by Intel on 5/25/2023. Configuration details are available in the addendum section of the whitepaper titled “Enhancing Performance and Power Efficiency of a Cloud-Native Broadband Network Gateway (BNG) Using 4th Gen Intel® Xeon® Scalable Processors”