Tech Mahindra and AWS Collaborate to Accelerate Open Radio Access Network Deployments with Amazon EKS Anywhere, Leveraging Tech Mahindra’s netOps.ai Platform

Open Radio Access Network (O-RAN or Open RAN) is revolutionizing the telecom landscape by offering flexibility, innovation, and intelligence through an open and disaggregated architecture. The cloudification of RAN functions introduces stringent requirements to meet real-time processing, time synchronization across the nodes, and optimal networking to achieve desired performance. The O-DU (O-RAN distributed unit) function is the most computationally intensive and time-critical component in an O-RAN system, requiring deployment closer to the radio cell site. This necessity drives the need to manage Kubernetes clusters on-premises or at the far edge. The Amazon EKS Anywhere provides a far-edge on-premises platform to meet the real-time processing requirements of O-RAN workloads.

The blog outlines the essentials of edge cloud deployment for O-DU network function on EKS Anywhere, emphasizing the importance of comprehensive testing to ensure its performance and success.

Importance Of Cloud In Open Ran

The O-DU performs functions at the RLC, MAC, and physical layers of the 3GPP RAN stack, which are computationally complex and latency-sensitive. These functions directly influence wireless channel throughput and coverage, impacting the user experience. Thus, the O-DU function must be hosted at the edge locations or radio cell sites at the far edge.

The O-DU can run on Amazon EKS Anywhere, a solution designed for telcos who want to retain the existing commercial-off-the-shelf (COTS) hardware while building an end-to-end platform and management framework from cell sites to an AWS Region cloud. As a Container-as-a-Service (CaaS) layer, EKS Anywhere can operate on bare metal servers, managing clusters in telcos' on-premises data centers while leveraging AWS-supported Kubernetes services.

EKS Anywhere offers two deployment architecture options:

1. Self-managed standalone clusters
2. Management/workload clusters

In the O-RAN deployments, the management/workload cluster architecture is widely used due to the highly distributed nature of the O-RAN and the need to manage numerous edge sites. EKS Anywhere management clusters are typically deployed in central or regional data centers. 

These clusters are responsible for the lifecycle management of the remote workload clusters. This hierarchical architecture simplifies operational tasks, enabling efficient management of multiple remote workload clusters in a typical O-RAN environment.

EKS Anywhere deployment topologies

EKS Anywhere can support O-RAN workloads that run on EKS Anywhere workload clusters and also accommodate customer-selected servers and designated hardware components such as L1 accelerators and network interface cards. These components enable Precision Time Protocol (PTP) support and low-latency access to the devices with a single root I/O virtualization container network interface (SR-IOV CNI) and data plane development kit (DPDK). Using the EKS console, telcos can view all the Kubernetes clusters running on cell sites, AWS outposts, local zones, and regions. This capability provides a unified management framework for deploying 5G CNFs, from O-DU to 5G core network functions.

Challenges

O-RAN introduces a new level of complexity by utilizing open interfaces and involving multiple vendors across various layers. Unlike traditional single-vendor RAN systems, multi-vendor O-RAN systems must ensure seamless interoperability between different modules, conformance of open interfaces, and end-to-end integration across the technology stack, testing, and validation. Achieving these goals requires extensive testing to ensure the O-RAN system meets performance, reliability, and security standards.

Choosing a suitable CaaS platform and O-RAN NF validation on the CaaS platform is a crucial step in building high-performance 5G networks. There are key considerations for cloud platforms to provide real-time performance, synchronization, high-performance networking, and scalability to handle the performance metrics of O-RAN systems. Telcos face challenges in developing an industrialized framework for testing and validating cloud platforms against these requirements. These challenges of CaaS platform testing and validation can be categorized broadly as follows: 

• Interoperability and integration
• Deployment at scale
• Performance optimization

Interoperability and Integration

In O-RAN networks, interoperability and integration challenges at the CaaS platform layer relate to ensuring that various software and hardware components from different vendors can work together effectively. To assure that multi-vendor solutions meet expectations on performance, time-to-market, scale, and quality, the platform needs to be tested and validated extensively across the hardware and software stack along with respective life-cycle management of multi-vendor modules of NF, servers, and K8s.

Deployment at Scale

Adapting cloud deployment at the far edge, with the scale of radio sites, remains challenging. The CaaS platform should support deployment at the far edge to accommodate thousands of radio sites (scalability), handle multiple simultaneous operations and workloads (concurrency), scale NFs (up or down), build resiliency, etc. Remote deployment, scalability, management, and observability are critical to validate the CaaS platform's ability to efficiently manage O-RAN networks. 

Performance Optimization

The CaaS platform must deliver high-performance networking and real-time processing and be able to abstract hardware capabilities to support the efficient operation of O-RAN networks. It should ensure reliable access to underlying resources (CPU, memory, and storage) at far-edge locations with constraints on space, power, and cooling. Leveraging technologies like DPDK / SR-IOV for data plane networking enables a time-sensitive environment for optimal functionality. 

Tech Mahindra + AWS Approach

Tech Mahindra and AWS have collaborated to test the EKS Anywhere platform for O-RAN integration and testing functions. This partnership leverages the power of both organizations for wider industry benefits and industrialization of the testing framework.

AWS brings deep cloud expertise to the EKS Anywhere platform, while Tech Mahindra, with its deep telecom industry expertise, knowledge, and experience of large-scale O-RAN deployments, tested and validated the platform.

The O-Cloud platform testing must be holistic and cover all aspects to ensure the platform is interoperable, scalable, secure, and resilient to meet performance metrics. The approach to O-RAN integration and testing of the CaaS platform includes these steps:

• Assessment and Planning
• Deployment and Integration
• System Testing and Validation
• Continuous Monitoring and Optimization

Approach to O-RAN integration and testing

Assessment and Planning

In this phase, we assess the requirement and evaluation of the infrastructure to map it with the goal or objective of the testing and validation activity. 

1. Requirement Analysis: We start by conducting a thorough assessment of the O-RAN requirements, including hardware, software, and network needs. This involves understanding the specific O-RAN components (CU, DU, and RU) and their interoperability requirements.
2. Infrastructure Evaluation: We evaluate the existing on-premises infrastructure to ensure compatibility with EKS Anywhere and O-RAN workloads. It also includes assessing network connectivity, hardware capabilities, and storage requirements.

Deployment and Integration

In this phase, the setup is created and deployed along with network functions (O-RAN NF workloads) using automation and orchestration tools.

1. Cluster Setup: We deploy Amazon EKS Anywhere clusters in the on-premises environment, configuring the cluster to meet the specific needs of O-RAN workloads. This involves setting up control plane and worker nodes, configuring networking, and ensuring high availability and scalability.
2. O-RAN Component Integration: The next step involves deploying O-RAN components as containerized applications within the EKS Anywhere clusters. Tech Mahindra integrates these components using its platform netOps.ai to ensure seamless operations.
3. Integration with netOps.ai: We integrate netOps.ai with the EKS Anywhere environment to leverage AI-driven capabilities. This integration provides enhanced visibility, analytics, and automation for deploying O-RAN workloads efficiently.

System Testing and Validation

In this phase, testing is conducted to assess key areas, such as functionality, performance, compliance, and security.

1. Functional Testing: We perform extensive functional testing using the available tools to ensure that all O-RAN components work together as expected. This includes validating the interoperability of CU, DU, and RU components and ensuring they meet performance and functionality requirements.
2. Performance Testing: This involves conducting performance testing using the available tools to evaluate the scalability and efficiency of the O-RAN deployment. This includes simulating various network conditions and traffic loads to ensure the system can handle real-world scenarios.
3. Compliance and Security Testing: We ensure the O-RAN deployment complies with industry standards and best practices. This includes testing for data protection, network security, and adherence to regulatory requirements.

Continuous Monitoring and Optimization

In this phase, the platform is integrated with observability tools and validated for real-time analytics and monitoring

1. Observability Tools: We integrate observability tools like Prometheus for monitoring and Grafana for visualization to provide real-time insights into the performance and health of O-RAN workloads.
2. Automated Scaling: We configure automated scaling policies within EKS Anywhere to ensure that the O-RAN deployment can efficiently adapt to changing workloads and demands.
3. Continuous Support: We also provide continuous support and maintenance to address any issues, apply updates, and optimize the performance of the O-RAN deployment.

Tech Mahindra has tested the EKS Anywhere platform across the areas below to validate it for O-RAN networks and workloads.

1. LCM
   • NF LCM (Network Function Lifecycle Management)
   • K8s Lifecycle Management (LCM)
2. Networking and Performance
   • Networking
   • Kubernetes
3. Scalability and Resiliency
   • Scalability
   • Concurrency
   • Failover, Backup
4. Observability and Security
   • Observability
   • Security and Authentication

EKS Anywhere platform testing and validation areas

Importance Of Automation In Open Ran Deployments

With the multi-vendor systems, different deployment scenarios, openness in a horizontal and vertical stack, volumetric scale of radio sites, and individual component life-cycle roadmap, the O-RAN network needs a holistic approach to automation and life-cycle management. Testing and validation of the cloud platform in the O-RAN network is a crucial step in achieving a robust, efficient, reliable, and resilient platform. Automated testing and deployment reduce the efforts required for lab validation testing, O-Cloud infrastructure deployment, and network function (NF) deployment, accelerating the entire process and reducing human errors and repetitive tasks.

Tech Mahindra strongly believes that automation and continuous integration, continuous delivery/deployment, and continuous testing (CI-CD-CT) are key elements of network deployments and should be an integral part of the solution framework for telcos. The Tech Mahindra netOps.ai platform provides a CI-CD-CT framework that can continuously integrate, deploy, and test the modules making it an automated workflow across multiple NFs and services.

Introducing CI/CD in the process can significantly improve efficiency and streamline the integration of various components into one cohesive system. The O-RAN network is expected to have a continuous pipeline of new hardware, software, and features in a CI/CD environment, where new releases will be continuous and require CI-CD-CT in both the lab and production environments.

netOps.ai CI-CD-CT flow

Conclusion

A comprehensive platform testing approach is essential to ensure the platform's reliability, performance, and compatibility with network function workloads. Testing across the areas of K8s LCM, NF LCM, networking, security, concurrency, Kubernetes functionality, observability, management, and other general aspects can validate the EKS Anywhere platform's readiness for production network deployment.

The O-RAN integration and testing must be conducted with the right set of tools and expertise to achieve seamless and efficient deployment and reduce time to market. Tech Mahindra platform NetOps.ai framework helps service providers to automate testing procedures, deployment, and life-cycle management with CI/CD pipeline and can accelerate the journey of service providers to deploy Open Radio Access Network on Amazon EKS Anywhere.