Question 1

Your organization is deploying a new InfiniBand-based storage network for real-time data processing. Ensuring high performance is critical, but at the same time, the network must be highly reliable to handle the lossless transmission of large datasets. The environment will scale over time, and both storage and compute traffic will share the same fabric. You must implement a solution that ensures high performance, scalability, and minimal packet loss under heavy loads. Which best practice is most effective for ensuring reliability and performance in a scalable InfiniBand deployment that handles both compute and storage traffic?

Options :

A : Implement QoS (Quality of Service) to prioritize storage traffic while using Virtual Lanes (VLs) to segregate traffic classes.

B : Separate storage and compute traffic into different physical networks to avoid contention and improve reliability.

C : Enable congestion control and adaptive routing while maintaining high MTU sizes across all nodes.

D : Utilize static routing and disable congestion control to ensure that paths are deterministic for critical data flows.

Answer: C

Question 2

Which of the following correctly describes the role of the HCA in InfiniBand’s connection-oriented (reliable) transport mechanism?

Options :

A : HCAs handle flow control and congestion management at the application layer, ensuring no data is lost in transmission.

B : HCAs perform hardware-based packet retransmission at the switch level to ensure reliability in connection-oriented transfers.

C : HCAs are responsible for managing reliable connections through queue pairs (QPs), which enable reliable data transfer between nodes.

D : HCAs manage connection setup and teardown for RDMA, ensuring that each connection is uniquely identified by a Local Identifier (LID).

Answer: C

Question 3

In InfiniBand, Virtual Lanes (VLs) are used to provide logical isolation of traffic within a single physical connection. What is the maximum number of Virtual Lanes supported by the InfiniBand standard, and what is their primary purpose?

Options :

A : 4 Virtual Lanes, used for low-latency traffic

B : 8 Virtual Lanes, used for data traffic prioritization only

C : 16 Virtual Lanes, with all used for data transmission

D : 15 Virtual Lanes, with one reserved for management traffic

Answer: D

Question 4

Which two of the following statements correctly describe InfiniBand’s subnet management and addressing standards as per the InfiniBand architecture? (Select two)

Options :

A : InfiniBand requires static IP addressing for managing device communication.

B : The InfiniBand Subnet Manager (SM) is responsible for configuring Local Identifiers (LIDs) and ensuring route calculation within a subnet.

C : InfiniBand devices use Ethernet MAC addresses to identify themselves within the InfiniBand fabric.

D : InfiniBand uses DNS services for name resolution within the network.

E : Global Identifiers (GIDs) are used in InfiniBand for routing between different subnets.

Answer: B,E

Question 5

In AI/ML training clusters, how does Remote Direct Memory Access (RDMA) over InfiniBand accelerate the performance of distributed deep learning?

Options :

A : RDMA over InfiniBand enables deep learning models to share memory across GPUs, allowing for larger models to be trained without memory bottlenecks.

B : RDMA over InfiniBand simplifies the orchestration of distributed training jobs by automating workload balancing across GPUs and nodes.

C : RDMA over InfiniBand automatically compresses the data being transferred, which reduces the bandwidth required for communication between GPUs.

D : RDMA over InfiniBand allows direct data transfer between GPUs without passing through the host CPU, significantly reducing the overhead during parameter updates.

Answer: D