NVIDIA has been pioneering the gaming industry since 1999. Since then, it has redefined modern computer graphics and revolutionized data center processing. More recently, with its GPU-based systems, the company ushered in modern artificial intelligence – a new era of computing – with the GPU acting as the intelligence for many devices.
The NVIDIA Networking division was formed in 2020, following the merger of acquired companies: Mellanox Technologies and Cumulus Networks. The combined expertise, experience and technology of the three entities gives users a choice of data center components that are increasingly faster and software-defined. It also provides a larger offering for increasingly popular AI, Cloud and HPC systems. Acquired Mellanox Technologies is a leading provider of end-to-end Ethernet and InfiniBand solutions for server, storage and hyperconverged infrastructure environments. The company offers a wide range of networking products of the highest performance and quality, including network cards, switches and dedicated software.
Cumulus Networks on the other hand has over 2,000 customers using 130 different hardware platforms running Cumulus Linux, the operating system for network switches. Together with Mellanox Technologies, they developed the Open Ethernet strategy in 2013, in which they are still technology leaders today. This strategy leverages their extensive knowledge and experience in building scalable data center solutions. Devices based on Open Ethernet technology are designed to be:
- Open – enabling the combination of third-party software with the raw device through open interfaces. This solution translates into freedom of choice and a high level of innovation. Mellanox Open Ethernet products allow for Software Defined Networking(SDN) deployment, management, monitoring and coordination via standard APIs and plug-ins. This enables the user to create efficient and unified solutions.
- Standard – based on well-known and widely used Layer 2 and Layer 3 protocols, while maintaining full compatibility.
- Linearly scalable – maintaining high performance, low cost and ease of management in even very large scale networks consisting of thousands of nodes.
- Efficient – Design optimized for the highest speed and port density, low latency and energy efficiency.
- Service oriented and virtualization – ensuring infrastructure monitoring and enforcing SLAs (Service Level