The latest 3rd Generation Intel Xeon Scalable processors, announced earlier this month, are delivering support for the PCIe 4.0 interfaces. To bolster severs based on the new Xeon Scalable processors, Intel has announced a new Ethernet network adapter.
Known as the Intel Ethernet Network Adapter E810-2CQDA2, the new adapter delivers up to 200 gigabits per second (Gbps) of bandwidth in a PCIe 4.0-compliant server. Each QSFP28 port supports up to 100Gbps, providing the throughput and functionality of two 100Gbps adapters in a single PCIe 4.0 slot.
Intel set out to match the bus (PCIe 4.0) with the bandwidth (Ethernet network speed). This new adapter, by taking advantage of the increased bandwidth, balances platform performance and helps ensure the network isn’t a bottleneck. It’s all about getting a balanced platform.
PCIe for speed
Support for PCIe 4.0 is important. This latest iteration of the PCIe interface basically doubles overall throughput, when compared with 3.0.
Your data-center customers need that bandwidth to support today’s data-centric computing use cases. These include accelerated network storage, cloud, communications and high-performance computing (HPC).
Intel’s new network adapter is available from tech distributors, resellers and other channel partners. Also, several hardware OEMs are bringing to market servers based on the 3rd Gen Intel Xeon Scalable processors. Some of these OEMs plan to support as many as 6 of the new Intel network adapters in a single server, bringing a physical server’s maximum throughput to 1,200 Gbps.
From the perspective of both the server OS and the user, the Intel Ethernet Network Adapter E810-2CQDA2 appears as 8 physical network adapters. That also means it can handle 8 separate networks.
800 Series support
Like other members of the Intel Ethernet 800 Series Network Adapter family, the new E810-2CQDA2 supports several key technologies from Intel. These include:
> Application Device Queues: For predictability at scale, ADQ enables application-specific data steering, signaling, and rate limiting using an optimized application thread to the device data path. This increases performance, reduces latency and improves throughput.
> Ethernet Port Configuration Tool: EPCT lets the user program each port to act as a separate physical adapter. There are many possible configurations, including 2 x 100Gb, 4 x 50Gb, and 8 x 25Gb.
> Dynamic Device Personalization: DDP improves packet-processing efficiency. The Intel Ethernet 800 Series Network Adapter firmware loads an enhanced DDP profile with many workload-specific protocols at driver initialization for greater flexibility.
> Remote Direct Memory Access: RDMA enables 2 machines on the same network to exchange memory without involving the operating system or the processor, further reducing latency. Both iWARP and RoCE v2 are supported simultaneously.
Security is provided using several methods. These include Zero Trust, which essentially means every server request, whether from inside or outside the network, must be verified to determine whether access should be granted.
Also, all Intel Ethernet 800 Series Network Adapter components comply with the NIST Cybersecurity Framework. Basically, this means that hardware-based security features protect the firmware and critical device settings.
These features include corruption detection and automated device recovery. If a device is attacked or corrupted, these features enable it to return to its originally programmed state.
Tell your data-center customers about Intel’s latest server processors and network adapters. Together, these components are powering the new generation of servers your customers need today.