As a result of a joint testing effort between 6WIND, NVIDIA, and VMware, a new performance white paper is now available. The test was conducted by using 6WIND Virtual Security Gateway (vSecGW) as a network function to show the performance advantages of VMware Telco Cloud Platform powered by VMware vSphere 8 and NVIDIA BlueField data processing units (DPU).
In 2022, VMware introduced the capability to run vSphere on DPUs, which enables vSphere 8 to offload NSX networking and security services onto a DPU such as NVIDIA BlueField.
One of the key benefits of the offloading is to free up CPU resources on the host, which can then be used to run larger and less latency-sensitive workloads separately from the performance-intensive network processing tasks performed by the DPU. Combining CPU and DPU resources on a single platform enables you to achieve much higher application performance. Such workload consolidation lets you deploy smaller server hardware in your data centers while gaining performance, which in turn not only reduces both CapEx and OpEx but also decreases power consumption.
Scalable and Aggregated IPsec Performance for Telecom Workloads
The purpose of the joint testing was to show how the 6WIND vSecGW can achieve scalable and aggregated (both upstream and downstream) IPsec performance of at least 25Gbps in a network by leveraging DPU hardware offloading in a realistic 4G/5G environment.
The above graph illustrates one of the test results for symmetrical, unencrypted, and IPSec-enabled traffic running through a single vSecGW instance.
The testing described in this white paper validates that the 6WIND vSecGW, running on Telco Cloud Platform, can efficiently and effectively support IPsec traffic using VMware vSphere with hardware-accelerated networking offloads, supported by the NVIDIA BlueField-2 DPU. The test results illustrate that higher throughput and faster application response times can be achieved regardless of the packet size with an IPsec.
Reducing Costs and Power Consumption with DPU Offloading
Offloading performance-intensive tasks, such as networking, to the DPU also frees up CPU cores that can be repurposed to support business-critical applications. With the offloading, more efficient practices can be applied to servers and data centers to reduce power consumption and costs.
In a world calling for faster packet processing, more cloud-native applications, and energy conservation, DPU offloading will become a necessity for both improving application performance and reducing TCO in the data center.
Learn more in the white paper: Accelerating Telecom Network Performance with DPU Offloading.
