Blog de Red Hat
Containers, software-defined wide area networking, cloud edge communications, and white-box devices
Containerized Software Defined Wide Area Networking (SD-WAN) on white-box x86 platforms offers a higher degree of flexibility and similar performance when compared to dedicated bare-metal solutions — especially when the goal is to deliver dynamic edge compute solutions with greater security capabilities.
Container technology enables software workloads to be deployed anywhere, allowing telecommunications service providers and others to deploy flexible SD-WAN networks that enhance the security profiles of multi-cloud communications and edge cloud communications. Example deployments could connect corporate applications hosted in multiple public and private clouds with corresponding edge compute devices or uCPE at branch offices, weather-proof nodes connected to IoT devices, or 4G/5G edge devices in fleet vehicles, trucks, heavy equipment, or vessels.
White-box technology makes edge compute devices easier to procure and support, reducing expenditure outlays for spare parts and recurring maintenance costs, while container technologies make deploying or upgrading software in edge devices simpler and easier.
Performance testing containerized SD-WAN: A multi-vendor configuration
Over the fall of 2021 and into early 2022, Intel, Red Hat, Turnium and TietoEvry collaborated on lab-based tests designed to show real-world, practical results on the benefits of containerized SD-WAN using Turnium’s SD-WAN Cloud Native Network Function (CNF) certified on Red Hat OpenShift, running on Intel-powered servers. TietoEvry provided the third-party testing environment, traffic simulators and technical staff to deploy, manage and test this software stack.
The goal of the evaluation was to compare the performance of dedicated, fixed-function deployments versus software-defined, containerized deployments.
The evaluation compared the performance of two edge device configurations:
A cloud-native containerized deployment using OpenShift running on an x86-based Universal CPE (uCPE) with Turnium’s SD-WAN deployed as a containerized application deployed, controlled and managed by OpenShift.
A bare metal deployment in which Turnium’s SD-WAN software image was directly installed on top of the uCPE operating system.
These edge nodes, or uCPE, communicated with core nodes via Turnium Aggregator software, running in:
A simulated public/private cloud offsite from the edge nodes, connecting the aggregators and uCPE using a public overlay network.
A simulated local aggregator positioned within the same local, private network as the edge nodes or uCPE.
Test results and concluding remarks
The results of the performance testing demonstrated that the containerized SD-WAN delivered the required performance while also delivering greater scalability, flexibility, ease of deployment and automation for edge compute solutions at scale.
For the duration of the test, it was easy to add new nodes to the network automatically. Automated network programming and capabilities, including wireline and wireless bonding and failover, enabled a survivable network to be turned up quickly. Deploying OpenShift made it clear that deploying, upgrading, or changing workloads was a simple task in managing many edge devices at scale, and decoupling the software from the hardware utilizing commercial off-the-shelf (COTS) hardware made it easy to size the device for core and edge nodes.
In terms of performance, tests indicated that OpenShift had a marginal impact on system load, with an approximate 2% additional CPU demand compared to running the SD-WAN implementation on bare metal. This low level marginal CPU impact provides room to grow and run additional containerized workloads within the uCPE, enabling businesses and enterprises to benefit from edge compute and public or private 5G deployments.
By coupling computing at the edge with technologies such as Turnium’s containerized SD-WAN platform and Red Hat OpenShift, it becomes possible to deploy workloads or applications easily to take advantage of the favorable cost and computing power made available by x86 COTS hardware.
To download the complete report and view test results, please explore more here.
About the author
John has extensive experience guiding organizations to success via innovative architecture of products, digital strategy, and customer and user experiences - particularly in online and mobile marketing, e-commerce, product management, web analytics and interactive content development.