Canonical and Spectro Cloud have collaborated to develop an effective telco edge cloud solution, Cloud Native Execution Platform (CNEP). CNEP is built with Canonical’s open source infrastructure solutions and Spectro Cloud’s Palette containers-as-a-service (CaaS) platform. This technology stack empowers operators to benefit from the cost optimisation and agility improvements delivered by edge clouds in a highly secure and performant way.

Now the solution has been shipped to the Intel Xeon platforms. When run on 4th Gen Intel® Xeon® Scalable processors, CNEP brings real-time processing capabilities to telco clouds, with support for real-time Ubuntu and the acceleration provided by Intel® Xeon® Scalable processors that is necessary to achieve real-time processing at the hardware level. This is a successful outcome of the recent close collaboration between Intel and Canonical engineering teams. CNEP with real-time processing capabilities can now be delivered as a solution to the telco market.

We are going to showcase this solution at MWC Barcelona 2024 , book a meeting with us to check out the demo.

Telco edge clouds

With the softwarisation of network services and the adoption of cloud networking in the telco sector, the architecture of mobile networks has evolved significantly. Modern telecom networks are no longer run by all-in-one systems deployed on a central location. Instead, operators can scale their systems and offer their services closer to users, thanks to highly scalable, distributed and cloud-native architectures.

Telco operators are increasingly deploying cloud computing systems at the edge of their networks, which are often referred to as edge clouds. According to the IDC spending guide forecast published in February 2023, service providers will invest more than $44 billion in enabling edge offerings in 2023. This trend has emerged due to the change in infrastructure architecture and the evolution of mobile networking software which is now based on components that run on containers as microservices.

Edge computing is predicted to grow even more, as the technology has brought efficiency, flexibility and scalability to telecom systems in deployment and operation. STL partner’s revenue forecast notes a prediction of $445bn in global demand for edge computing services in 2030.

Five key requirements for edge cloud success in telco

To unlock the benefits of cloud computing, operators need an effective infrastructure stack to host cloud-native software on edge clouds. Telco deployments are highly demanding, and so a suitable infrastructure stack should satisfy these five key requirements:

Autonomous operations

It is critical to minimise operational maintenance for edge clouds. These clouds are large in number, and it is costly to maintain systems manually, especially when they are deployed close to radio equipment where it is impractical for administrators to visit deployment sites physically. The solution is to ensure that edge clouds can be operated in an autonomous manner.

Secure

Telco networks are part of our critical infrastructure, carrying sensitive user data. Systems must comply with all necessary security standards and have hardening measures to safeguard user information.

Minimal but variable in size

A minimal footprint is one of the defining characteristics of an edge cloud. A few server hardware nodes may be all that is needed to set up a small cloud that would run a number of cell sites. That being said, there is no single size solution – requirements may change based on what an operator intends to run at its edge network. Therefore, infrastructure must be able to scale as and when needed.

Energy efficient

A telco operator typically runs a large number of sites for its radio networks. Even a 2% reduction in energy consumption translates to significant cost savings. This means that the ideal edge cloud solution must be optimised at every layer of its stack and have features that support running and operating only what is needed with no extras. It should also support advanced hardware and software features to reduce power consumption.

Highly performant

Telco networks must deliver user data quickly and reliably – service quality and reliability depends on it. Solutions at the telco edge must support the latest technology and enhanced features that enable faster delivery of information at every layer of the hardware and software stack. One of these performance requirements is support for mission-critical business applications and time-sensitive compute at the telco edge, requiring support for ultra-reliable and low-latency processing at both software and hardware layers of the complete cloud stack.

Challenges

Edge clouds need a software stack that is built with multiple virtualisation technologies, which makes it challenging to integrate and set up a fully functional system. Addressing the five requirements mentioned above with modern open source cloud technologies is a complex task. Despite the clear benefits those technologies bring, there are still gaps to fill.

Maintaining updates and upgrades in a cloud system is of paramount importance for smooth system operation while ensuring system integrity and security. However, a typical distributed telecom system deployment has many edge sites each running a virtualisation infrastructure. Furthermore, both the virtualisation software and the application workloads that run on a cloud environment have a large set of dependencies. Given this scale and complexity, it is simply not feasible to manually perform updates and upgrades to maintain these systems.

Besides updates and upgrades, operational procedures such as deployment, scaling and runtime maintenance, are highly repetitive across all telco edge cloud sites. Without a scalable system, it is not possible to operate a telco-edge infrastructure in a cost-efficient way.

CNEP

Cloud Native Execution Platform (CNEP) addresses the five key requirements of successful edge clouds when deploying and maintaining their distributed telco cloud infrastructure. It offers a software stack that is efficient, secure, performant and modular.

The solution stack is tailored for the needs of telco edge clouds from bare metal to containers. It consists of Canonical’s Metal-as-a-Service (MAAS) and Canonical Kubernetes solutions that together deliver the bare metal performance and orchestration required by the telecom sector while enabling the flexibility and agility of cloud native environments. Integrated with Spectro Cloud’s Palette , the solution provides automation for deployment of Canonical’s cloud native edge cloud stack at scale at multiple edge sites.

CNEP can offer real-time compute capabilities to telco cloud operators. This is the result of the advanced technology collaboration between Intel and Canonical. This close collaboration has produced a highly innovative integration solution for fast and ultra-reliable processing at telco edge clouds. It is a combination of complementary technologies at the silicon and operating system levels. Together, Intel’s cutting-edge silicon and the Ubuntu real-time kernel provide a robust stack from hardware to software which can address the needs of modern radio access networks, such as Open RAN, as well as mission-critical business applications running as multi-access edge computing (MEC) services at edge clouds.

Platform features

Cloud Native Execution Platform simplifies onboarding, deployment and management of Kubernetes clusters. Canonical Kubernetes is a light-weight, zero-ops, and purely upstream CNCF certified Kubernetes distribution by Canonical, with high availability, automatic updates and streamlined upgrades. It is the container orchestrator in CNEP, tailored for telco edge clouds, with optimised performance, scalability, reliability, power efficiency and security.

CNEP offers an array of features that make it ideally suited to telco use cases.

Multi-site automation

CNEP provides multi-site control, observability, governance and orchestration with zero-downtime upgrades. Through Spectro Cloud Palette, operators can seamlessly deploy, configure and manage all their telco edge clouds from a central location.

Palette not only manages bare metal automation and provisioning with MAAS but also achieves deployment and management of Kubernetes clusters, all through

Cluster API (CAPI) . It gives operators rich and fine-grained control over their Day 2 operations, such as patching and configuration changes. The platform also provides full observability and role based access control (RBAC) capabilities.

Repeatable deployments

In CNEP, operators can achieve repeatable and reliable Kubernetes cluster deployments with automation at scale using Palette across multiple geographical sites. With Palette, CNEP achieves decentralised policy enforcement and self-healing for autonomy and resilience at scale. This provides operators with a consistent end-to-end declarative management experience.

Self-healing by Palette in CNEP is achieved by continuously monitoring the state of the deployed Kubernetes cluster at each site and comparing it against the desired cluster state. Any deviation between the two states is addressed by bringing the cluster to the desired state based on policies.

Cloud native, reliable and software defined

CNEP is cloud native and reliable for containerised workloads. Kubernetes supports Cluster API to meet the complex needs of highly distributed edge node onboarding, secure deployment and substrate provisioning. It also supports all popular container networking interfaces (CNI), including Cilium, Calico and Flannel, as well as Kube-OVN as a CNI for software defined networking.

For management and control of object, block and file storage, Kubernetes integrates with

Canonical Charmed Ceph , which is a flexible software-defined storage controller solution. CNEP provides support for these CNIs and Charmed Ceph out of the box.

Automated hardware at scale

Bare metal hardware provisioning with MAAS enables operators to automate their edge hardware infrastructure, and gain visibility and control over their hardware resources. This provides agility in system deployment with full automation in configuration and operating system deployment.

MAAS supports CAPI to enable hardware automation operations while deploying and managing Kubernetes clusters. With Palette, CNEP achieves bare metal automation at scale across multiple edge cloud sites through MAAS CAPI.

Secure and compliant

Ubuntu Pro provides security compliance, hardening and auditing, as well as support to the edge cloud infrastructure as a whole and to the cloud native telco workloads running in containers. It provides security patches, hardening profiles, standards compliance and automated CVE patches for an extensive set of open source packages (over 23000). CNEP supports multiple security standards. For instance, both Ubuntu Pro and Palette have conformance to FIPS 140-2.

As CNEP’s container-confined execution environment, effectively isolating it fromr orchestrator, Kubernetes security is mission-critical, and our solution ensures that it is safeguarded. In addition to the security features of Ubuntu Pro, Canonical Kubernetes runs in a snap, which is a change in the host system and other software running on the host. This provides a sandbox environment and protects the container orchestration environment from external threats.

The attack surface is reduced as much as possible to minimise entry points to the platform and protect it from malicious attempts. This is achieved by the opinionated design of Kubernetes, chiselled container images and Ubuntu Core.

Canonical Kubernetes has a minimal footprint that includes all necessary components but nothing extra. It is easily extensible with its modular structure as needed. Similarly, chiselled container images include only the packages needed to execute your business applications, without any additional operating system packages or libraries. In constrained environments, Ubuntu has a minimal flavour - Ubuntu Core. This provides operators with an immutable operational environment where the system runs on containerised snaps.

Besides the security features provided by Canonical's telco edge cloud stack at each telco site, Spectro Cloud Palette brings additional security capabilities to CNEP. This includes native security scanning for the full deployment stack, conformance scans, and penetration testing. Palette provides further patching and monitoring capabilities, along with the role-based access control offered as part of CNEP.

Performant

CNEP is highly-performant across the telco infrastructure stack.

At the container orchestration level, Canonical Kubernetes supports the latest enhanced platform features that streamline packet delivery between containerised applications and external services. It supports technologies such as GPU acceleration and CPU-pinning.

At the operating system and silicon hardware levels, Ubuntu Pro and Intel silicon together bring real-time compute capabilities that meet the stringent requirements of delay-sensitive telco applications and the networking stack. This enables low latency and ultra-reliable communications, which means applications can communicate with users and devices with the fastest possible performance at the OS level.

Ubuntu’s real-time kernel accompanies Intel Time Coordinated Computing (TCC) and FlexRAN reference software by prioritising real-time tasks at the software level, and ensures their execution is uninterrupted, delivering minimal response time. It provides the necessary bounded time delay guarantees in the operating system to mission-critical applications powered by FlexRAN applications.

Next generation Intel CPUs with Intel TCC deliver optimised computing performance with low latency, supporting IEEE 802.1 Time Sensitive Networking (TSN). TCC prioritises real-time workloads at the hardware level when accessing cache, memory, and networking resources. With support for TSN, Intel TCC ensures timeliness through synchronisation across network nodes. Intel CPUs with FlexRAN reference software integrate advanced signal processing, hardware acceleration and highly efficient input/output (I/O) modules for real-time performance.

CNEP runs on bare metal hardware, which makes it ideal for efficiency at the telco edge. Automatic updates provided by Ubuntu Pro’s kernel Livepatch service gives an uninterrupted environment to telco workloads and the networking stack.

Cost-efficient

CNEP is designed to be efficient with minimal energy consumption at the telco edge.

Canonical Kubernetes is modular and can be extensible as necessary; it comes with a sensible set of default modules in place. This enables Kubernetes to be more efficient with the best possible use of system resources.

Ubuntu Core has the same properties. It is minimal, with services running on snaps, providing a small footprint which consumes much less resources without sacrificing performance.

MAAS enables significant cost reductions on two aspects thanks to its hardware automation capabilities. Indeed, on one hand, MAAS automates OS provisioning and software deployment on bare metal hardware - reducing operational costs and human errors. On another hand, IT administrators can optimise hardware utilisation based on workload conditions managed by MAAS.

Those automation capabilities are augmented by the multi-site automation capabilities brought by Palette. CNEP achieves cost savings in terms of simplified deployment and management of the edge infrastructure, as engineers no longer need to physically visit deployment sites.

Summary

Powered by Canonical’s industry-leading open source infrastructure solutions, Intel’s cutting-edge 4th Intel® Xeon® Scalable processors, and with multi-site automation provided by Spectro Cloud Palette, CNEP can seamlessly scale across distributed telco infrastructure. It is ideal for cloud native telco workloads, edge computing business applications, and mobile networking stack, such as Open RAN CU/DU/RU and distributed 5G user plane. The solution is secure by design thanks to Ubuntu Pro, and highly efficient with support for real-time kernel and other enhanced platform features.

Get in touch

Canonical provides a full stack for your telecom infrastructure. To learn more about our telco solutions, visit our webpage at ubuntu.com/telco or get in touch .

Book a meeting with us at MWC Barcelona 2024 .

Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries.

Learn more

Fast and reliable telco edge clouds with Intel FlexRAN and Real-time Ubuntu for 5G URLLC scenarios

Telco edge cloud automation at scale

Reducing latency at telco edge clouds with Ubuntu real-time kernel

Safeguarding your telco infrastructure with Ubuntu Pro

How to build carrier-grade infrastructure using enterprise open source solutions

On-demand webinar: Kubernetes on bare metal: ready for prime time!