High-throughput hardware-accelerated elastic network DPU CONFLUX®-2200E | 消费电子产品 | 零件,组件和原材料 | 计算机与外设
High-throughput hardware-accelerated elastic network DPU CONFLUX®-2200E
中国内地
|
生产线所在地:中国内地
展位编号: 3E-D03 | 创意科技及智慧城市 |
产品规格
型号编号

2200E

品牌

CONFLUX

应用

Network offloading scenario;Virtualization scenario;Cloud-native network offloading scenario;Custom service logic scenario.

产品简介

CONFLUX®-2200E series is a SmartNIC product based on K2-Pro, a programmable network ASIC self-developed by YUSUR technology. The network acceleration engine embedded in K2-Pro supports hardware virtualization, network hardware offloading and acceleration. This product supports OpenFlow protocol and P4 programming language, and meets the extreme requirements of high throughput and low latency in various application scenarios. This product can be widely used in cloud computing, internet, finance, telecommunications, government,  enterprise and other fields. 


【Highlights】 

Self-supporting and controllable R&D

Fully self-developed high performance ASIC, KPU architecture with proprietary intellectual property rights, and numerous patented technologies


High-performance network offloading and acceleration

OVS hardware offloading with hardware connection tracking capability, significantly improve forwarding performance and reduce the load of CPU. In addition,  stateless hardware offloading functions are supported, such as checksum and segmentation.


Cloud-native network offloading

Deeply adapted to mainstream CNIs, including ovn-kubernetes and Kube-OVN. The dedicated network acceleration engine can offload and accelerate the cloud native network, effectively improve the overall performance and reduce the load of host CPU.


P4 language programming

The ability of programming on heterogeneous platform, gives great  help to the rapid development and verification of custom protocols, and it is possible to define the data plane forwarding process according to specific service requirements.


Virtualization

SR-IOV hardware virtualization is supported, and light-weighted PCIe network interfaces can greatly reduce the load of host CPU, improving network performance and reducing forwarding latency.


Domestic ecosystem support

Adapted to the mainstream CPU architectures, operating systems and some domestic ones, building the most complete ecosystem in the industry.