NCP-AII Question 51
Single answerYou are designing the network topology for a data center that will support large-scale AI workloads requiring high bandwidth and low latency between compute nodes. Which of the following network topologies would be the most suitable for this purpose?
- A
Fat-tree topology
- B
Ring topology
- C
Star topology
- D
Bus topology
Show answer and explanation
Correct answer: A
Explanation
The fat-tree topology is commonly used in modern data centers because it addresses the high bandwidth and low latency requirements of AI and high-performance computing workloads. It ensures efficient communication between nodes by leveraging a hierarchical structure with multiple paths, which reduces congestion and enhances fault tolerance.
- A. Correct.
Fat-tree topology is ideal for data centers supporting high-performance AI workloads because it provides high bandwidth, scalability, and redundancy. It minimizes bottlenecks and ensures efficient communication between compute nodes.
- B. Incorrect.
Ring topology is not suitable for large-scale AI workloads as it suffers from latency issues and limited scalability. Data must traverse multiple nodes to reach the destination, which can slow communication.
- C. Incorrect.
Star topology is less effective in large-scale AI infrastructure since it creates a single point of failure at the central hub and does not scale well for systems requiring high bandwidth.
- D. Incorrect.
Bus topology is outdated and unsuitable for high-performance computing or AI workloads due to its limited bandwidth and scalability. Communication is bottlenecked along a single shared medium.