Telco-grade Sylva-compliant Canonical platforms

In December 2023, Canonical joined the Sylva project of Linux Foundation Europe to provide fully open-source and upstream telco platform solutions to the project. Sylva aims to tackle the fragmentation in telco cloud technologies and the vendor lock-in caused by proprietary platform solutions, by defining a common validation software framework for telco core and edge clouds. This framework captures the latest set of technical requirements from operators when running telco software workloads as cloud native functions (CNF), such as 5G core microservices and Open RAN software.

Sylva’s mission is to support 5G actors in their efforts to drive convergence of cloud technologies in the telco industry – taking into account interoperability across 5G components, TCO with open source software, compliance with regulations and adherence to high security standards. CNFs from vendor companies can then be operated and validated on reference implementations of the cloud software framework defined by Sylva. 

To test and validate telco vendor CNFs, Sylva has deployed cloud-native platforms based on a multi-deployment model as Kubernetes (K8s) clusters on bare metal or OpenStack. These CNFs often require telco-grade enhanced platform features like SR-IOV, DPDK, NUMA, and Hugepages, along with support for a range of container networking interfaces (CNI). In this blog, we explain how Canonical’s Sylva-compliant infrastructure solutions satisfy these requirements.

Canonical’s open source platform solutions for Sylva

Canonical’s product portfolio is closely aligned with Sylva’s objectives and strategies. It provides a variety of features that Sylva aims to include in the latest modern telecom infrastructure deployments. The project has already deployed validation platforms running on Ubuntu, and also leverages hardened Ubuntu 22.04 images.

Canonical Kubernetes is a CNCF conformant enterprise-grade Kubernetes with high-availability. It delivers the latest pure upstream Kubernetes, which has been fully tested across a variety of cloud platforms of all form factors, including provisioned bare metal systems, Equinix Metal and OpenStack, and architectures including x86, ARM, IBM POWER and IBM Z. It supports the Cluster API (CAPI), which is mandated by Sylva to provision Kubernetes. With CAPI, an operator can update Kubernetes clusters through rolling upgrades without disruption and initialise their workloads. 

For telco edge clouds, Canonical Kubernetes can scale as a lightweight Kubernetes solution with self-healing, high-availability and easy clustering properties. This provides a minimal footprint for more energy-efficient operations at edge clouds. It can equivalently scale up at regional and central clouds where a larger footprint is needed in a data centre. 

Based on Canonical Kubernetes, Canonical’s Cloud Native Execution Platform (CNEP) aligns with the Sylva platform features and architectural design. With CNEP, Kubernetes clusters are offered to telco operators on bare metal hardware, where hardware provisioning and cluster operations can both be controlled and orchestrated via Cluster API centrally. 

CNEP’s set of supported features makes it ideal for operators who want to adopt a Sylva compliant platform with validated telco CNFs from vendors, e.g. 5G core and Open RAN as well as MEC CNFs, such as content delivery networking (CDN) software. The platform software stack fully supports the Sylva design from bare metal to containers, with capabilities including:

• Bare metal provisioning operations automated via Cluster API
• Enhanced platform awareness features, such as SR-IOV, DPDK, CPU pinning, Hugepages and NUMA
• Ubuntu operating system with CIS security hardening, compliant with FIPS, NIST 800-53, PCI DSS, DISA STIG, ISO 270001 standards
• A real-time kernel for mission-critical applications and latency-sensitive telco workloads, such as Open RAN DU and 5G UPF
• Fully upstream and CNCF-compliant Canonical Kubernetes that provides operators with an industry-standard and production-grade Kubernetes container orchestration platform with multi-tenancy features, exposing Cluster API
• A wide range of CNIs, required by vendor CNFs and the Sylva validation framework, such as Cilium, Calico, Multus, and others
• Ceph as a backbone for distributed multi-tenant storage with configurable data protection and encryption
• Full observability, with support for the Canonical Observability Stack, consisting of popular open source software tools Grafana, Prometheus, and Loki, supporting logging, monitoring and alerting
• Role based access control (RBAC) features at platform, Kubernetes and bare metal provisioning levels

In addition to Canonical Kubernetes and our CNEP solution, Canonical OpenStack supports the advanced platform features that Sylva validation platforms need, including SR-IOV, DPDK, CPU-pinning, NUMA, Hugepages, PCI passthrough, and NVIDIA GPUs with virtualisation. It has native support for both Ceph and Cinder as storage components, both of which are included in the Sylva platform design and roadmap.

About the Sylva project 

Aligned with telco operator needs, Sylva envisions cloud-native telco software execution on Kubernetes platforms. Operators look to deploy Kubernetes clusters at their telco edge, regional and core clouds, providing them with a uniform cloud-native execution environment.

Modern telco infrastructure is distributed, deployed across multiple locations with tens of thousands of far-edge clouds, thousands of near-edge clouds and tens of regional clouds. This calls for deploying and managing a large number of Kubernetes workload clusters at geographically dispersed locations, controlled by management cluster(s) located at regional and central clouds. To tackle this challenge, Sylva has defined a software framework for telecom software platforms based on Kubernetes that are deployed on a large scale. 

Modern telco clouds must also support a set of enhanced platform features often required by telco CNFs. Towards this, the project’s validation platforms verify that (i) the deployment platform supports the requirements of a CNF in test, and (ii) the CNF can correctly deploy on the platform and successfully consume these platform features.

Kubernetes cluster management

Sylva follows a declarative approach with a GitOps framework to manage a high volume of physical nodes and Kubernetes clusters. Infrastructure lifecycle management covers Day 0 (build and deploy), Day 1(run), Day 2(operate) operations, with fault management, updates and upgrades. The project provides automation with CI/CD pipelines where a set of scripts produce and maintain Helm charts that include Kubernetes deployment and operational resource definitions. 

A dedicated work group, called Telco Cloud Stack, has developed tooling for cluster deployment and lifecycle management (LCM). This tooling is based on the Flux GitOps tool, which keeps clusters and infrastructure components in sync with their definitions in Git repositories. 

To manage the Kubernetes clusters and bare metal provisioning with this tool-chain, Sylva leverages Cluster API (CAPI).

Validation of telco CNFs on Sylva platforms

CNFs from different vendors are validated on Sylva platforms for the interoperability between the CNFs and the platforms. The project’s validation program ensures that telco operators who deploy platforms with software components that follow the Sylva reference implementations gain two benefits: (i) verified telco CNF functionality on their cloud platforms, and (ii) verified support for the telco-grade platform features which these CNFs require.

The project has a dedicated work group called the Sylva Validation Center, which tests deployment of vendor CNFs on the project’s validation platforms, where Kubernetes runs on either bare metal hardware or on OpenStack. 

The validation of a CNF under test on a Sylva platform starts with identifying the necessary set of platform capabilities that the CNF requires, including CNIs, and then installing and configuring the platform with those capabilities. Once the platform has been configured, a first set of smoke tests are run to verify the platform’s support for this set of features. Once the CNF has been deployed on the platform, some functional tests are performed to verify that the deployment is correctly done, and all the necessary Kubernetes pods are healthy in ready state. Finally, operators may run additional tests on CNFs if deemed necessary.

Canonical’s open source software and solutions meet the platform feature requirements by telco CNFs as tested by the Sylva Validation Center, such as SR-IOV, Multus CNI, and Real-time Linux. Validating telco CNFs on Canonical’s platforms for Sylva will also ensure that our platforms with support for these advanced features are verified by Sylva to run these CNFs.

Sylva platform roadmap

In its roadmap for 2024, project Sylva is planning to add support for new features in its validation platforms, such as near real-time Linux, immutable operating system for far-edge clouds and GPU offloads. Canonical’s software platforms follow Sylva’s vision and have support for these features already today, with Real-time Ubuntu, Ubuntu Core immutable OS, support for precision time protocol (PTP) and more.

Canonical is committed to making Sylva a benchmark platform for executing telco network functions. This commitment entails Canonical’s contribution to the infrastructure-as-code scripts that compose Sylva, to enable our open source solutions for Sylva, and to align with the evolving technical scope of the project.

Summary

Linux Foundation Europe’s Sylva project has defined a platform architecture for validating cloud-native telco network functions on Kubernetes. This provides telco operators with guidance on how to achieve a uniform cloud infrastructure, covering edge, regional and central cloud locations, ultimately aiming at multiple objectives, including cost reduction, interoperability, automation, compliance and security.

The project emphasises the central role of open source platforms with standard and open APIs, which brings a modular approach when designing and deploying telco cloud systems. 

Canonical offers fully upstream and telco-grade open source solutions that align with the Sylva platform architecture, including Canonical Kubernetes and Canonical OpenStack. We also engineered an innovative platform solution, CNEP, which is fully inline with the Sylva visions on multi-tenancy, multi-site Kubernetes clusters,  bare metal with full automation of hardware provisioning and cluster lifecycle management performed over industry-standard Cluster API.

Contact us

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

Further reading

Canonical joins the Sylva project

Bringing automation to telco edge clouds at scale

Canonical Kubernetes 1.29 is now generally available

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