NCP-AII Question 53
Single answerA data center administrator is designing a network architecture for an AI workload that requires low latency, high bandwidth, and scalability to support GPU-accelerated workloads. Which network topology is the most suitable for this use case?
- A
Spine-Leaf
- B
Ring
- C
Bus
- D
Star
Show answer and explanation
Correct answer: A
Explanation
The Spine-Leaf topology is the most suitable for AI and GPU-accelerated workloads in data centers due to its ability to deliver low latency, high bandwidth, and scalability. It ensures equal access to resources across the network, which is critical for high-performance computing tasks.
- A. Correct.
The Spine-Leaf topology is specifically designed for modern data centers and is optimal for AI workloads. It provides high bandwidth, low latency, and scalability by ensuring consistent performance across all nodes.
- B. Incorrect.
The Ring topology is not commonly used in modern data centers. It introduces latency and scalability challenges as the number of nodes increases.
- C. Incorrect.
The Bus topology is outdated and unsuitable for modern data centers, especially for high-performance workloads, as it creates a bottleneck with a single shared communication channel.
- D. Incorrect.
The Star topology, while simple, lacks scalability and is prone to failure at the central node, making it unsuitable for AI workloads in data centers.