Adapter Fault Tolerance

Adapter Fault Tolerance (AFT) is a simple, effective, and fail-safe approach to increase the reliability of server connections. AFT gives you the ability to set up link recovery to the server adapter in case of a cable, port, or network interface card failure. By assigning two server adapters as a team, AFT enables you to maintain uninterrupted network performance.

AFT is implemented with two server adapters: a primary adapter and a backup, or secondary, adapter. During normal operation, the backup will have transmit disabled. If the link to the primary adapter fails, the link to the backup adapter automatically takes over.

Preferred Primary Adapter

With multiple adapters installed, you can specify one as the Preferred Primary adapter. For example if you have a server with an IntelPRO/100 Intelligent Server adapter as the primary adapter and an IntelPRO/1000 adapter as the secondary, you would want the PRO/100 Intelligent Server adapter to be the preferred primary. In this scenario, if the PRO/100 Intelligent Server adapter fails, the PRO/1000 will take over. Then when the PRO/100 Intelligent Server adapter is replaced, it will automatically revert to being the primary adapter in the team.

If a Preferred Primary is not selected, the IntelPROSet II will attempt to select the best adapter based on adapter model and speed.

Mixed Adapter Teaming

AFT supports up to four PRO/1000 or PRO/100 adapters per team, in any mix.

Adaptive Load Balancing

Adaptive Load Balancing (ALB) is a simple and efficient way to increase your server’s transmit throughput. With ALB you group server adapters in teams to provide an increased transmit rate (up to 400 Mbps) using a maximum of four adapters. The ALB software continuously analyzes transmit loading on each adapter and balances the rate across the adapters as needed. Adapter teams configured for ALB also provide the benefits of AFT. Receive rates remain at 100 Mbps.

To use ALB, you must have two, three, or four server adapters installed in your server or workstation and linked to the same network switch.

16

Intel Server Board SHG2 Product Guide

Page 16
Image 16
Intel SHG2 manual Adaptive Load Balancing, Adapter Fault Tolerance

SHG2 specifications

Intel SHG2, short for Server High-Performance GPU 2, represents a significant evolution in graphics technology designed to meet the demanding requirements of data centers and high-performance computing applications. Developed to support a range of workloads, SHG2 is particularly optimized for artificial intelligence, machine learning, and data analytics.

One of the defining features of the Intel SHG2 is its impressive processing power. With numerous cores capable of parallel processing, it can execute multiple tasks simultaneously, leading to faster computation times. This architecture makes it incredibly efficient for applications that require heavy calculations and real-time data processing.

The SHG2 also embraces advanced fabrication technologies, moving to smaller nodes that enhance power efficiency while delivering higher performance. These innovations not only allow for improved thermal management but also contribute to the reduction of overall energy consumption, making the SHG2 a more environmentally friendly choice for server operations.

In terms of memory support, the SHG2 offers a robust configuration, accommodating high-bandwidth memory (HBM) that ensures quick access to critical data. This feature is essential for applications needing rapid data retrieval and integration, such as deep learning models where every millisecond counts.

Moreover, the SHG2 is built with compatibility in mind. It integrates seamlessly with existing Intel architectures, allowing organizations to leverage their current investments while upgrading their processing capabilities. This compatibility makes it easier for businesses to enhance their infrastructure without undergoing a complete overhaul of their systems.

Another salient characteristic of the Intel SHG2 is its support for virtualization. This allows multiple workloads to be handled simultaneously on a single GPU, maximizing resource utilization and enhancing overall system performance. As enterprises increasingly adopt cloud-based solutions, the ability to virtualize GPU resources becomes critical.

Security is also a major consideration in the design of the SHG2. Intel has incorporated various security features to ensure that data is protected against unauthorized access and vulnerabilities that could threaten system integrity.

As businesses increasingly turn to AI and machine learning to drive innovation, the Intel SHG2 positions itself as an essential tool for those looking to harness the power of advanced computing. With its impressive features, powerful performance, and forward-thinking design, is a strategic choice for organizations aiming to enhance their computational capabilities and meet the challenges of tomorrow's technological landscape.