Question 1

Which of the following descriptions of distributed training is wrong?

Options :

A : In data parallelism, each computing node trains the complete model based on local data

B : Distributed training can support the training of massive data and complex models

C : Data parallelism and model parallelism are common distributed training methods

D :

In model parallelism, each compute node will only interact with the trained sub-model and obtain the final model through model aggregation

Answer: D

Question 2

Which of the following data center deployment scenarios is typically used to achieve business continuity and ensure uninterrupted service to users in the event of a failure?

Options :

A : Intra-city active-active

B : Two places and three centers

C : Edge DC

D : Hybrid cloud

Answer: A,B,D

Question 3

Which of the following descriptions of open source OpenStack storage components is correct?

Options :

A : Cinder is generally used for long-term storage of persistent type of data at rest

B : Cinder provides object storage services

C : Swift provides block storage services

D : Swift is generally used to store virtual machine images, data backups, etc

Answer: D

Question 4

Which of the following is not part of the analysis scope of the iMaster NCE-FabricInsight network snapshot function?

Options :

A : BGP and OSPF protocols

B : Device startup configuration

C : Number of packets lost on the interface

D : ARP entries

Answer: B

Question 5

In Huawei's CloudFabric solution, if you want to use a third-party VAS device to provide value-added services, which of the following VAS device management modes should you use if the VAS device cannot be managed by the controller ? ?

Options :

A : Service Policy

B : Network Policy

C : Service Manager

D : Network Manager

Answer: B