NCP-AII Question 47
Select 3You are tasked with setting up the network fabric ports for hosts in an NVIDIA AI infrastructure environment to ensure optimal connectivity and performance for data-intensive workflows. Which of the following configurations should you prioritize during the setup process?
- A
Ensure each port is configured for consistent MTU (Maximum Transmission Unit) across all hosts in the network.
- B
Enable RDMA (Remote Direct Memory Access) capabilities on supported ports for reduced latency and improved throughput.
- C
Disable VLAN tagging to simplify network configurations for hosts.
- D
Verify the ports are configured to support the correct Ethernet or InfiniBand protocol based on the network fabric type.
- E
Configure all ports to operate at the lowest available speed to ensure compatibility across devices.
Show answer and explanation
Correct answers: A, B, D
Explanation
When setting up network fabric ports for hosts in NVIDIA AI infrastructure, it is essential to focus on configurations that optimize performance and compatibility. Consistent MTU settings, enabling RDMA, and ensuring the correct protocol are critical steps. These configurations collectively enhance data transmission efficiency and support the demanding requirements of AI workflows.
- A. Correct.
Correct. Consistent MTU configuration across all hosts is crucial to avoid packet fragmentation and ensure smooth data transmission in high-performance environments.
- B. Correct.
Correct. RDMA reduces CPU overhead and latency, making it essential for accelerating AI workloads and improving data transfer efficiency.
- C. Incorrect.
Incorrect. Disabling VLAN tagging can lead to poor network segmentation and security issues. VLANs are often used to isolate traffic in AI infrastructure environments.
- D. Correct.
Correct. Configuring the correct protocol (Ethernet or InfiniBand) ensures compatibility with the network fabric and optimizes performance.
- E. Incorrect.
Incorrect. Operating at the lowest available speed undermines the performance requirements of AI workloads, which typically demand high-speed connections.