Network Interface Cards

Product Description

For cloud and enterprise architects building modern data centers, the Alveo U25N provides a comprehensive 25GbE SmartNIC platform that brings true convergence of network and security acceleration functions like OVS and IPsec into a single platform.

The U25N SmartNIC platform is based on the fusion of three technologies, a powerful System on a Chip (SoC) that includes an FPGA and multi-core Arm processor, and a proven XtremeScale™ X2 Ethernet Controller. The FPGA enables in-line hardware acceleration and offload with maximum efficiency while avoiding unnecessary data movements. 

The Arm cores handle exception traffic processing, and FPGA-related management and statistics. The X2 Ethernet Controller chip provides a platform for processing two ports of 25 Gigabit Ethernet via field-proven software drivers

Key Features & Benefits | Powerful SmartNIC

The U25N delivers ultra-high throughput, small packet performance, and low latency. The host interface supports standard NIC drivers as well as Onload® kernel bypass to provide both TCP and packet-based APIs for network application acceleration. The U25N is also tamper-resistant and provides secure flow monitoring with enforcement via a stateless firewall.

Programmable Platform

The U25N SmartNIC contains a programmable FPGA processing all the network flows. Each flow can be individually delivered to the host and/or streamed in hardware through bump-in-the-wire network acceleration functions. The platform also enables the ability for customers to accelerate custom workloads.

Hardware Accelerated for Clouds

Cloud service providers are deploying SmartNIC fabrics to achieve virtual switching and micro-segmentation of services that scale linearly with CPU cores and network links. The U25N is a platform for the industry's first converged SmartNIC fabric, including shrink-wrapped applications.

AMD ADVANTAGES

• Bump-in-the-wire model to offload and accelerate networking and security functions
• Turnkey advanced network accelerations including Open vSwitch (OvS), IPsec, and firewall
• Application customization and programmability enables flexibility to add custom workloads
• Baseline NIC feature parity with field-proven XtremeScale™ Ethernet controllers

The XtremeScale™ series of Ethernet network adapters are designed for high-performance electronic trading environments and enterprise data centers; including artificial intelligence, big data, analytics, hyperscale, machine learning, storage, and telco applications.

With record-breaking low latency and high throughput, XtremeScale network adapters provide real-time packet and flow information to thousands of virtual NICs. This combination of ultra-high bandwidth, ultra-low latency, ultra-scale connectivity, and packet telemetry allows X2 series adapters to scale with each server, virtual machine or container.

For cloud data centers, Onload® application acceleration software dramatically accelerates and scales network-intensive workloads such as in-memory databases, software load balancers, and web servers. With Onload, data centers can support 4X or more users on their cloud network while delivering improved reliability, enhanced quality of service (QoS), and a higher return on investment, without modification to existing applications.

In electronic trading environments, Onload kernel bypass application acceleration and DPDK services deliver superior small packet performance with sub-microsecond hardware latency.

In addition, XtremeScale adapters support the precision time protocol (PTP) fabric service for apps that require synchronized time stamping of packets down to single-digit nanosecond resolution.

Designed for Enterprise, Private and Public Cloud Datacenters

Target Applications

• Electronic trading platforms
• Public, private and hybrid-cloud data centers
• Bare metal, containers, and virtualized environments
• Content delivery networks

Key Features & Benefits

• Sub-microsecond latency
• Near-zero jitter
• Highly scalable
• Predictable, deterministic system performance
• Industry's highest message rates
• Precision Time Stamping for accurate synchronization
• NVMe™ /TCP storage ready
• SolarCapture® Pro network performance monitoring

The new XtremeScale™ SFN8000 adapter family provides Solarflare's best performance yet, with higher bandwidth, lower latency and higher packet rate. The SFN8000 family adapters are based on a common hardware platform, enabling customers to configure the features and performance they want to meet the particular needs of each application. Capabilities such as ultra-low latency, Onload® kernel bypass, class leading clock synchronization accuracy providing MIFID II readiness, ServerLock® Filter Engine, SolarCapture® Pro and more can be enabled to run on any SFN8000 adapter on the network, making this the one Ethernet adapter to deploy across your data center or cloud infrastructure.

CPU-Efficient Networking
The XtremeScale SFN8542 dual-port 40G Ethernet QSFP+ server adapter delivers faster, more efficient processing of network traffic to accelerate a wide range of applications. The SFN8542 has 16 lanes of PCIe 3.1 making it an ideal workhorse for 40GbE data center and cloud workloads, providing high-bandwidth, high packet rate and CPU-efficient processing of network traffic. The SFN8542 can be configured as a dual-port 40GbE or a quad-port 10GbE adapter as one QSFP+ port can also operate as four 10GbE SFP+ ports.

Configurable Software
When configured with Solarflare software licenses, the SFN8542 can address ultra-low latency applications, highly accurate clock synchronization and hardware time stamping, packet filtering, packet capture and more. Adapters can be purchased pre-configured with popular software bundles, or configured on-site with software options.

Ultra-Low Latency
The platform's hardware optimizes packet flow to and from the user application. When licensed for ultra-low latency, it delivers unrivaled sub-microsecond TCP and UDP latency. With Onload, the industry's best application-to-application performance is enabled via Solarflare's unique and patented kernel bypass techniques.

Accelerating Applications
The SFN8542 is ideal for scale-out cloud, web and CDN application environments. Solarflare adapters are used in a wide range of use cases, including Software-defined networking (SDN), Network functions virtualization (NFV), web content optimization, DNS acceleration, web firewalls, load balancing, NOSQL databases, caching tiers (Memcached), web proxies, video streaming and storage networks. The SFN8542 also provides the ideal link between fast NVMe storage and the network. In addition when used with Onload, the SFN8542 further reduces host processing overhead and accelerates web and cloud traffic. When used with ServerLock Filter Engine, the SFN8542 provides an additional layer of defense against DDoS attacks at the network ingress point to the server. Attacks can be absorbed without degradation of "good" traffic, enabling a 3 to 4x improvement in packet-level filtering server headroom.

Scalable, High-Performance Virtualization
Solarflare is the proven leader in VM density and performance, and the SFN8542 raises the bar with 2048 vNICs, SR-IOV, overlay network acceleration e.g. VXLAN, NVGRE. In addition, the SFN8542 supports kernel bypass running in a guest VM, so users can run ultra-low latency Onload or SolarCapture Pro in a virtualized or cloud environment.

The SFN8542 is also fully flow-aware, providing flexible layer 2-4 flow processing and flow steering natively. The benefits are efficient flow acceleration in almost every environment from the data center to the stock exchange.

Xilinx's XtremeScale™ X2522 10/25G dual-port Ethernet network adapters are designed for high-performance electronic trading environments, and cloud and enterprise data centers; including artificial intelligence, big data, analytics, machine learning, storage, and telco applications.

NETWORKING
Record-breaking low latency and high throughput network adapters provide real-time packet and flow information to thousands of virtual NICs. This combination of ultra-high bandwidth, ultra-low latency, ultra-scale connectivity, and packet telemetry allows X2 series adapters to scale with each server, virtual machine or container.

ACCELERATION
For electronic trading environments, Onload® kernel bypass application acceleration and DPDK services deliver superior small packet performance with sub-microsecond hardware latency.

Onload dramatically accelerates and scales network-intensive workloads such as in-memory databases, software load balancers, and web servers. With Onload, data centers can support 4X or more users on their cloud network while delivering improved reliability, enhanced quality of service (QoS), and a higher return on investment, without modification to existing applications.

In addition, X2522 adapters support the precision time protocol (PTP) fabric service for apps that require synchronized time stamping of packets down to single-digit nanosecond resolution.

TELEMETRY
SolarCapture® Pro packet capture software enables X2522 adapters to capture and time stamp packets in hardware, persist packet to disk, and replay packets streams. This feature provides the essential ingredient for corporate data centers to run applications such as high-precision network performance monitoring, analysis, and regulatory compliance.

WORKLOADS
The X2522 adapters are ideal for scale-out cloud, web, and CDN application environments. Use cases include softwaredefined networking (SDN), network functions virtualization (NFV), web content optimization, DNS acceleration, web firewalls, load balancing, NoSQL databases, caching tiers (Memcached), web proxies, video streaming, and storage networks.

X2522 adapters speed up TCP, unicast, multicast and UDP traffic for a wide variety of trading applications. From missioncritical exchange gateways and matching engines on the financial exchange side, to feed handlers, order routing, algorithmic trading engines, messaging applications, order execution, data distribution and client communications on the trading side, XtremeScale network adapters are considered the de facto standard in electronic trading environments.

The PLUS variant of the Flareon® Ultra SFN8542 dual-port 40GbE QSFP+ server I/O adapter adds a suite of enabled software capabilities that provide out-of-the box support for ultra-low-latency Onload®-accelerated sockets as well as PPS and PTP time synchronization support.

The SFN8542-PLUS is capable of associating a hardware time-stamp with every incoming and outgoing Ethernet frame for application troubleshooting or performance monitoring. These features make the SFN8542-PLUS the ultra-low-latency Ethernet adapter of choice to deploy as a key component throughout Financial Trading infrastructures where either one or two ports of 40GbE or four ports of 10GbE are required.

CPU-Efficient Networking
The Flareon Ultra SFN8542-PLUS dual-port 40G Ethernet QSFP+ server adapter delivers faster, more efficient processing of network traffic to accelerate a wide range of applications. The SFN8542-PLUS has 16 lanes of PCIe 3.1 minimizing latency and maximizing bus throughput between the adapter and the host. The SFN8542-PLUS can also be configured as a quadport 10GbE adapter with one QSFP+ port operating as four 10GbE ports.

Flexibility in Operation
Configured either in 40GbE or 10GbE mode, in a virtualized environment the adapter supports hypervisor bypass via SRIOV providing virtualized guests with near baremetal network performance. Multiple physical functions are also available enabling the ports to be partitioned when connected to 802.1Q trunked switch ports with a logical interface per VLAN being presented to the operating system. Ports on the same adapter or across multiple adapters may be bonded for redundant operation. This redundancy support also extends to Onload® with accelerated sockets capable of running across both active-passive and active-active LACP-enabled links.

The SFN8542-PLUS provides the following licensed features enabled:

Ultra-Low Latency
The Flareon Ultra SFN8542 offers fast and efficient hardware that optimizes packet flow to and from the user application. The SFN8542-PLUS unlocks unrivaled sub-microsecond latency both at Layer 2 and TCP and UDP via Onload.

Onload
Onload is Solarflare's industry's leading fully RFC-compliant kernel-bypass user-space TCP/IP stack. Unrivaled applicationto-application TCP and UDP latency is available with no modification to the applications being accelerated. Applications may communicate via Onload to any other RFC-compliant TCP/IP stack with seamless interoperability.

Precision
Time The SFN8542 is equipped with a Stratum 3 oscillator providing class-leading clock stability. The SFN8542-PLUS enables the use of sfptpd, Solarflare's time synchronization daemon, which leverages the adapter's ability to apply hardware time-stamps to incoming and outgoing Ethernet frames to synchronize its oscillator either via the PTPv2 protocol or a combination of PPS and NTP. Sfptpd also enables the Linux real-time clock to be synchronized to the external timing reference via the adapter clock. Any time-stamps obtained from either packet events or via the Linux time API may now be correlated directly with the external timing source. These features allow users to implement event traceability that exceeds the MIFID II requirements.