MENU
A company is building a network fabric for an AI training cluster that will train large language models. The cluster includes 256 GPUs distributed across 32 servers. Security requirements mandate isolation between different training projects and allowing shared access to a central storage system.
Which approach provides the required security isolation and maintains optimal performance for GPU-to-GPU communication?
Suggested Answer: B
Get Premium 300-640 Questions
as Interactive Web-Based Practice Test or PDF
Real Exam Format
Detailed Explanations
Instant Access
Get Your Practice Test
5 comments · newest first
JO
Johnson
5 days ago
Option B seems solid for isolation and performance.
GI
Gilberto
10 days ago
Option D seems interesting with QoS, but I can't remember if it really addresses the isolation requirement effectively. I need to think more about how policy-based routing works in this context.
RI
Richelle
15 days ago
I lean towards option C since VLANs are a common solution for segmentation, but I worry about the complexity of managing ACLs for storage access.
IV
Ivette
20 days ago
I feel like option A is too costly with all those physical networks. I practiced a similar question where VLANs were used, but I can't recall if they provided enough isolation for security.
JU
Jutta
26 days ago
I think option B with VXLAN sounds familiar. I remember studying how it allows for better scalability and isolation, but I'm not entirely sure about the Group Policy part.