H13-629 HCIE-Storage (Written) Questions and Answers

pubs

master

msdb

tempdb

127.0.0.1 localhost129.22.221.200 cluster-scan129.22.221.201 node1129.22.221.211 node1-vip192.168.222.211 node1-priv129.22.221.202 node2129.22.221.212 node2-vip192.168.222.212 node2-priv

127.0.0.1 localhost129.22.221.200 cluster-scan129.22.221.201 node1129.22.221.211 node1-vip192.168.222.211 node1-priv129.22.221.202 node2129.22.221.212 node2-vip192.168.221.212 node2-priv

127.0.0.1 localhost129.22.222.200 cluster-scan129.22.221.201 node1129.22.221.211 node1-vip192.168.222.211 node1-priv129.22.221.202 node2129.22.221.212 node2-vip192.168.222.212 node2-priv

127.0.0.1 localhost129.22.221.200 cluster-scan129.22.222.201node1129.22.221.211 node1-vip192.168.222.211 node1-priv129.22.221.202 node2129.22.221.212 node2-vip192.168.222.212 node2-priv

True

False

The production system equipment network card occupancy rate is high.

The production system equipment network card occupancy rate is low.

The production system equipment CPU occupancy rate.

The production system equipment CPU occupancy rate is low.

The engineer thinks the installation PC is connected to NIC1 to complete the software deployment, so connect the installation PC to NIC1

Direct Deployment of Software Without Disconnecting Front End Network and All Network Nodes

Software Deployment Before starting, the engineer configures a temporary back-end IP address on the back end network port B3-0 of the 9000 node (connected to the installation PC)Department DHCP server

The installation of PC directly through the network cable connectedto the OM management network switch to complete the software deployment

True

False

Restore the entire data file

roll back to the Council before crying to a certain point in time

Recovery control file

open state can not be restored action

can only be opened when creating a file object, business can not be opened or closed during data deduplication and compression function

Using online processing to deduplicate data before it is written to disk

When bothdeduplication and compression are turned on, the data will be deduplicated and then compressed

Use smart cache can accelerate the search performance of fingerprints deleted

Dynamic domain name, which returns the dynamic IP address of the node for the client, supports the failover function and is recommended for CIFS protocol sharing

Static domain name, which returns the domain name of the node's static external IP address toclients, is recommended for NFS protocol sharing

dynamic external IP address, node failure, the dynamic IP address can be automatically assigned to other working nodes, the client can pass againThis IP address is connected to the OceanStor 9000

static external IP address, the node fails, the client can not access the oceanstor 9000 through the IP address

True

False

Agent-Agent monitors the resources of each system and provides the status to each machine's HAD

HAD Cluster service that monitors and manages the cluster's NAS

Under the management of the master node of the cluster, Gab sends the communication information between the cluster members to the point communication

LLT (Low-Latency Transport) sends information to all nodes in the cluster through a low-latency delivery protocol

oplocks

noguest

fs_mode

hide_unreadable

Scenario: Sequencing large IO services, especially 1M IO scenarios, with the phenomenon that the back-end read bandwidth is much larger than the front-end read bandwidthTuning: Turn offthe corresponding LUN prefetch policy

scene: the order of pressure reading and writing business conditions, high utilization of hard drivesTuning means: increase the unplug-thresh device hard disk device

Scenario: Oracle database in the operation ofdata backup, the data on the LUN bandwidth does not meet the requirements and accompanied by high utilization of hard diskTuning means: the formal creation of POOL to increase the depth of the points

Scenario: In the case of single concurrent or low concurrent read service, the delay is found to be largeTuning means: change disk unplug 8 Modify unplug - thresh increased from 4 to 8

Does not support the system disk backup and recovery operations

When enabling the snapshot-based backup mode, the NDMP backup feature requires the file system to support snapshots. Otherwise, the backup task can not be successfully performed.

Based on the snapshot backup, snapshots are created for the current directory. Because the extended attributes of the current directory receive the snapshot protection, the current directory extended attributes will be backed up.

Does not support the system time zone or other time-related adjustments, does not support daylight saving time.

vg01

vg02

vg01 vg02

empty

Check if the size of the migrated file is the same as the original data.

Check if the number of documents corresponds to the original data

Check if each file with read and write permissions corresponds to the original data

Whether the time for reviewing the case file corresponds to the original data.

Stop the current production business

Add the remote device to the original production array

With the disaster recovery array as the master LUN and the original production array as the slave LUN, create the remote replication relationship between the DR and the production end and wait for the initial synchronization to be completed.

Make an incremental synchronization.

Delete theoriginal production array consistency group.

The service plane uses GE channels for data transmission.

The service plane connects the host IO channel, disk IO channel and GE switching channel.

Plane redundancy between each other, when the business plane fails, allowing the use of control plane transmission services.

When the GE switch channel fails, you can use the PCIe switch channel as a backup data transmission channel.

True

False

SQL Server Authentication

Windows authentication

Log in as superuser

Other ways when landing

If the system has monitored the file X hasbeen cached on Node B, the client client1 can read the file directly from Node B.

If the system monitors that node B has cached file X, then node A can read the file from node

After HUAWEI OceanStor 9000 processes the request for reading the file X from client1, the system reserves the message on both node A and node B.

If the system monitors that node A has cached file X, then node A only caches the data of file X and does not cache its verification data

When the access to the N8500 is all on the intranet, and the client traffic is not equal, it is recommended to configure the "balance-alb" mode (that is, bond 6 mode).

When creating a LUN, write business for a long time, the stripe depth recommended 512KB.

If the RAID group is configured as RAID 6, the number of member disks is recommended to be 9 disks, and it is not recommended to configure the RAID group.

Fencing LUN should be placed in the same RAID group.

In the scenario where the actual RPO indicator requires zero, you can choose to synchronize remote replication.

Synchronous remote replication canbe used to map the main and secondary storage of the two sets of LUN mapping to the host to create a scenario volume.

Synchronous remote replication on the remote disaster recovery can be well done.

Because synchronous remote replication does not usesnapshots, after synchronous remote replication is interrupted abnormally, the data on the primary LUN needs to be resumed in reply.Start synchronization to the slave site.

After a snapshot of the LUN of the redundancy site is created, the snapshot is not activated atthe same time.

Disaster recovery storage side of the database resides in the cache has not been written to the hard disk.

When creating a snapshot on the disaster recovery end LUN, the production end is copying data to the disaster recovery end.

The production end and disaster recovery end adopt the disaster recovery solution of incremental LUN copy. The incremental data is abnormally interrupted during the synchronization process and has not been repaired.

4940G

4280G

5940G

5280G

Customer needs collection and refining

Backup capacity calculation

Project background research

Network environment is collected

fs create striped fs01 10g 3 RG1-LUN1, RG2-LUN1, RG3-LUN1 stripeunit = 2048 blksize = 8192

fs create mirrored fs02 10g 2 RG1-LUN2, RG2-LUN2 stripeunit =2048 blksize = 8192

fs create mirrored-stripe fs03 10g 2 2 RG1-LUn1, RG2-LUN1 protection = disk stripeunit = 2048 blksize = 8192

fs create striped-miiro fs04 20g 2 3 RG1-LUn1, RG1-LUN2, RG2-LUN1, RG2-LUN2, RG3-LUN1, RG3-LUN2 protection = diskstripeunit = 512 blksize = 1024

The mode of operation of the array can be in-controller load balancing or inter-controller load balancing.

Load Balancing Mode Including wheel training algorithm, the maximum queue depth algorithm, the maximum workload algorithm.

After the connection is interrupted between the LUN home controller and the host, the host accesses the LUN through the peer controller. In this case, the over-trail software defaults the switched home controller of the LUN to the remote controller.

When multipath detects a preferred controller failure recovery, the I / O is resumed immediately at the preferred controller by default.

Create a virtual snapshot of the slave LUN and activate it, then map the snapshot LUN to the host for authentication.

After the snapshot LUN is mapped to the host, the host can read and write the snapshot LUN.

After the verification is completed, stop the upper-layer service application, release the LUN mapping, and finally delete the snapshot.

Stop the snapshot before the host can write snapshot LUN.

Support for SMB3 multichannel features that take full advantage of client-side multi-core CPUs and bandwidth resources to dramatically improve performance.

When a hard drive has multiple objects at the same time to write requests, support merged into a large continuous IO, significantly improve system performance.

Internal file system replication, without having to go through the client directly in the system to complete high-speed

Supports 10GE and PCIE networking, providing high performance through high-speed inter-node interconnection and reducing system internal delay.

Recommended threshold 10GBSoftware threshold 15GBGrace time 10 daysHard threshold 20GB

Recommended threshold 10GBSoftware threshold 18GBGrace time 10 daysHard threshold 20GB

Recommended threshold10GBSoftware threshold 15GBGrace time 15 daysHard threshold 20GB

Recommended threshold 8GBSoftware threshold 15GBGrace time 10 daysHard threshold 20GB

Hard drive capacity

Metadata space size

Hard drive type

RAID type

True

False

You can plan back-end storage network for 9000 internal communications, without planning gateway address, and front-end network with the same network segment IP address.

Oceanstor 9000 front-end network and management network can be considered together, planning for the same network segment.

Back-end CE6850 switches need to use four stacked cables for stacking connection to ensure bandwidth requirements

The external NTPserver address and the Oceanstro 9000 front-end network need to communicate to ensure that the oacenstor 9000 system and client applications maintain time synchronization.

The service plane uses GE channels for data transmission

The service plane connects the host IO channel, disk IO channel and GE switch

Plane redundancy between each other, when the business planefails, allowing the use of control plane transmission of business data

When the GE switching channel is faulty, the PCIe switching channel can be used as the backup data transmission channel

offline mining

database

mailbox

public folder

True

False

For the "configuration operating system failed" error, check the node / var / log partition and clean up the partition, initialize the system, and then re-run the software deploy

If it is not a failure, restart the deployment after reinstalling the deploy deployment package

After the deployment fails, run the sh /opt/huawei/deploy/script/clean.sh command on each node to initialize the system and restart the software department

After deployment abnormalities, reboot all nodes, reinstall oceanstor tookit, and start the software deployment

Snapshot does not support nested, that is, the snapshot has been created directory, the parent directory and subdirectories can no longer create a snapshot

The root directory of the snapshot (that is, the directory used to create the snapshot) does not supportdeletion

Hard links are protected by snapshots, soft links are not protected by snapshots

For a snapshot with an expiration time set, the system automatically deletes the snapshot but does not back up the storage space occupied by thesnapshot data after the snapshot expires

True

False

use rmmod + driver name, remove directly

Directly uninstall the driver rpm package

kill directly kill the drive process

use modeprobe-r + driver name, remove directly

True

False

When a file is in the locked state, the file protection duration can be manually prolonged or shortened.

A locked file will shift to the expired state if the file system WORM clock value is greater than the file atime value.

After the WORM file system compliance clock has exceeded the file expiration time, an appending file will shift from the appending state to the expired state.

Only files of 0 byte can shift from the locked state to the appending state.

The size of the LUN provided by the VTL back-end storage does not exceed 4TB. The default configuration is write-back, mirroring, and the depth of the slice is 64KB

SIR data The number of LUNs is 2's power of N

SIR index LUN size is at least 4% of SIR data LUN size

It is not recommended to divide RAID across frames

1600 KB/s

400 KB/s,

15 MB/s

10 MB/s

Correct

mistakes

Backup tasks will consume a lot of LAN resources.

Backup tasks will consume few LAN resources.

Backup tasks will consume a lot of CPU resources of the production system.

Backup tasks will consume few CPU resources of the production system.

SAN

NAS

hot_add

NBD

With continuous random read: RAID 5 performance is superior to RIAD10

With sequential read and write, RAID 5performance are better than RAID10

With sequential write, RAID10 better than RAID5 performance

With continuous random read, RAID 5 and RAID 10 performance is equal.

control file

data file

log file

configuration file

True

False