Your browser is not supported. Please update it.

21 February 2020

Ericsson moves towards bare metal to boost 5G core performance

Ericsson has unveiled some of the new products and other news that it had intended to showcase at the cancelled Mobile World Congress next week. Among these is a cloud-native 5G core, an important step with the first deployments of 5G Standalone networks – which require a 5G core – expected this year.

Many operators will wait for several years before implementing 5G Standalone, waiting for the fledgling core technology to stabilize in performance and price. And some early adopters will stop short of a full cloud-native implementation and opt for some of the simpler virtualized architectures that are available. But for most the end goal is a fully cloud-based platform, implemented in containers and microservices on common cloud infrastructure. Those microservices can be assembled and re-assembled, Lego-style, to support new services, changing user demands and to centralize or distribute functionality as required.

The big mobile network vendors, plus Cisco and several smaller providers like Mavenir and Affirmed Networks, are readying their cloud-native cores for the first deployments in 2020. Some are more fully cloudified that others – Affirmed and Mavenir can even support deployments in the public cloud, though this option is likely to apply mainly to smaller operators or discrete platforms like those for the IoT; while on the other hand, Huawei is not yet supporting open container technologies and components like those from the Kubernetes community.

Ericsson’s latest updates include a dual-mode 5G core which is optimized for AI-based analytics. It is the latest vendor to say it now supports containers on bare metal infrastructure, rather than just within virtual machines on platforms like OpenStack (Huawei also announced this, for the user plane only). Ericsson says bare metal deployment can save operators up to 30% of their operational costs compared to virtual machines, and the container approach will enable automated deployment and distribution of core functions, and other 5G applications, across central and edge cloud locations. Eventually, the aim is to eliminate the virtualization layer altogether to improve hardware efficiency and overall performance.

The first announced customer for the new cloud-native 5G core is Telstra of Australia, which has a long-standing Ericsson-only procurement policy.

Ericsson also introduced cloud-based microservices for a range of enterprise, IoT and voice services, which can be deployed on existing network assets.

And the core will have in-built software probes to enable AI insights, working with a line of new or enhanced network services. These include Network Intelligence and Omni Network Channel, which use AI-based predictive analytics and automation to reduce network complexity even further than cloud-native alone, in order to drive down cost and improve time to deploy and optimize connectivity and applications.

Ericsson claims Network Intelligence can reduce critical incidents affecting network performance by up to 35% and automatically resolve issues within an average of five minutes of receiving the data.

Omni Network Channel is a “unified digital workspace” in which operators can work together with Ericsson to address network performance issues in real time.

The Swedish vendor is also promising up to 30% reduction in energy costs in the RAN, thanks to a new AI-powered energy management service, which aims to reduce the need for site visits. Ericsson claims the service can reduce energy-related network opex by 15%, site visits by the same percentage, and energy-related outages by 30%.

Finally, Ericsson boasted of achieving a download rate of 4.3Gbps to a 5G smartphone, the fastest rate publicly demonstrated so far, by aggregating 800 MHz of millimeter wave spectrum.