I85PREP
⌘ K
HCIE-Storage study materials
HCIE-Storage study resources

HCIE-Storage practice questions

One question at a time. Build your HCIE-Storage understanding.

First 50 of 200 questions. Each one lists the choices, the correct answer, and an explanation. The checker at the top of this page saves progress in this browser when it is available.

Continue in the practice tool above to load more questions, 50 at a time.

Question 1Storage (Expert-level track content, domains not enumerated in source)

In a Huawei all-flash array, which technology is primarily responsible for reducing write amplification before data reaches the SSD media?

  1. Inline deduplication and compression Correct answer.
  2. Snapshot rollback
  3. Read-ahead prefetching
  4. LUN copy

Explanation

Inline deduplication and compression remove redundant and compressible data in the write path before it is persisted to SSD, which reduces the volume of physical writes and therefore write amplification on the media. Prefetching, LUN copy, and snapshot rollback do not transform write payloads in this way.

Question 2Storage (Expert-level track content, domains not enumerated in source)

Which RAID scheme provides dual-disk fault tolerance while allowing a single large LUN to span many disks without the write penalty of RAID 6?

  1. RAID-TP Correct answer.
  2. RAID 5
  3. RAID 10
  4. RAID 6

Explanation

RAID-TP uses three independent parity blocks per stripe, tolerating three simultaneous disk failures while permitting large LUNs across many members. RAID 6 tolerates two failures but incurs double parity update overhead on every write, and RAID 5 tolerates only one.

Question 3Storage (Expert-level track content, domains not enumerated in source)

A customer wants a secondary site that can take over production automatically within seconds and keep RPO near zero over metropolitan distances. Which Huawei solution design fits best?

  1. Active-active stretched cluster across two data centers Correct answer.
  2. Periodic snapshot shipping to a third site
  3. Asynchronous remote replication with manual switchover
  4. Backup to tape with offsite vaulting

Explanation

An active-active stretched cluster keeps both sites serving I/O simultaneously with synchronous mirroring, so a site failure causes near-zero data loss and fast automatic takeover within metro latency limits. Asynchronous replication and snapshot shipping both leave an RPO window, and tape vaulting is far slower.

Question 4Storage (Expert-level track content, domains not enumerated in source)

What is the primary function of a host multipath driver in a SAN environment?

  1. Compressing blocks before they are written to disk
  2. Providing redundant logical paths to a LUN and load balancing I/O across them Correct answer.
  3. Encrypting data at rest on the storage array
  4. Deduplicating repeated blocks within a volume

Explanation

Multipath software presents multiple physical paths to the same LUN as a single logical device, so I/O continues if one path fails and can be balanced across the surviving paths. Encryption, compression, and deduplication are array-side data services, not host path management.

Question 5Storage (Expert-level track content, domains not enumerated in source)

An administrator configures a HyperMetro pair and needs a consistency group to guarantee that all member LUNs fail over to the same point in time. What must be true of the member LUNs?

  1. They must all be thin-provisioned.
  2. They must all use RAID 5.
  3. They must all belong to the same HyperMetro consistency group and share the same pair state. Correct answer.
  4. They must all reside on the same physical disk.

Explanation

A consistency group coordinates the write ordering and failover of member remote replication pairs so that all LUNs recover to a common point in time. Membership in the group and a consistent pair state are the requirements; physical placement, provisioning type, and RAID level are not.

Question 6Storage (Expert-level track content, domains not enumerated in source)

Which statement accurately describes the behavior of a thin-provisioned LUN when its capacity is exhausted and no automatic expansion is configured?

  1. The array silently steals capacity from other LUNs
  2. Existing data is deleted to free space
  3. The LUN automatically converts to thick provisioning
  4. Writes to the LUN fail or the LUN goes read-only once allocated space is exhausted Correct answer.

Explanation

Thin provisioning allocates physical space on demand, so if the storage pool cannot satisfy new allocations the host sees write failures or the LUN becomes read-only. The array does not take capacity from other LUNs, convert the LUN automatically, or delete data.

Question 7Storage (Expert-level track content, domains not enumerated in source)

What is the main purpose of a write-back cache with battery or capacitor backup on a storage controller?

  1. To compress data before it reaches the pool
  2. To reduce the cost of SSD media
  3. To acknowledge host writes from cache and flush them later while protecting against power loss Correct answer.
  4. To eliminate the need for RAID parity

Explanation

Write-back caching acknowledges writes once they are in protected cache, improving latency and throughput, and the battery or capacitor preserves that data for flush after power loss. Caching does not change media cost, replace parity, or perform compression.

Question 8Storage (Expert-level track content, domains not enumerated in source)

A database workload issues many small random writes. Which array-side setting most directly improves its performance?

  1. Enabling read-ahead prefetch
  2. Increasing the snapshot retention count
  3. Scheduling more frequent full backups
  4. Using write-back cache with SSD-backed pool tiers Correct answer.

Explanation

Small random writes benefit from absorbing writes in cache and landing them on low-latency SSD media, which reduces per-I/O latency. Prefetch helps sequential reads, snapshot retention affects capacity, and backups add load rather than improving write performance.

Question 9Storage (Expert-level track content, domains not enumerated in source)

In Huawei remote replication, what distinguishes asynchronous replication from synchronous replication?

  1. Asynchronous replication only works within a single data center
  2. Asynchronous replication requires no network link
  3. Asynchronous replication guarantees zero RPO
  4. Asynchronous replication acknowledges host writes before the secondary site confirms receipt Correct answer.

Explanation

Asynchronous replication returns write acknowledgement to the host once the primary has written locally, then ships data to the secondary later, which introduces an RPO window. Synchronous replication waits for secondary acknowledgement, providing zero RPO at the cost of latency.

Question 10Storage (Expert-level track content, domains not enumerated in source)

Which mechanism allows a storage array to present a point-in-time copy without duplicating all data immediately?

  1. LUN formatting
  2. RAID rebuild
  3. Full volume mirror
  4. Copy-on-write snapshot Correct answer.

Explanation

A copy-on-write snapshot preserves the original blocks by redirecting or copying them only when they are about to be overwritten, so the snapshot is created quickly with minimal initial space. Full mirroring duplicates all data, and rebuild and formatting are unrelated operations.

Question 11Storage (Expert-level track content, domains not enumerated in source)

An engineer plans to use SmartTier to optimize a mixed workload. Which data placement outcome should be expected?

  1. Only snapshot data is tiered
  2. All data is pinned to the highest-performance tier
  3. Data is evenly striped across all tiers regardless of access
  4. Hot data migrates to high-performance media and cold data migrates to lower-cost media Correct answer.

Explanation

SmartTier monitors access frequency and relocates frequently accessed extents to high-performance media while moving colder extents to capacity-oriented media. It does not pin everything to the top tier, stripe uniformly without regard to heat, or restrict tiering to snapshots.

Question 12Storage (Expert-level track content, domains not enumerated in source)

What is the role of an initiator IQN in an iSCSI storage network?

  1. It identifies the target portal group
  2. It uniquely identifies the host initiator for LUN masking Correct answer.
  3. It defines the RAID level of the LUN
  4. It sets the TCP window size

Explanation

The iSCSI Qualified Name uniquely identifies an initiator so that the target can map LUNs to the correct host through masking and access control. Target portal groups, RAID levels, and TCP tuning are separate concepts.

Question 13Storage (Expert-level track content, domains not enumerated in source)

A customer reports that a newly mapped LUN is visible in the array but not on the Linux host. Which troubleshooting step is most appropriate first?

  1. Replace the controller cache battery
  2. Delete and recreate the snapshot
  3. Rescan the host SCSI bus and verify LUN masking and host initiator registration Correct answer.
  4. Reinitialize the storage pool

Explanation

A LUN that exists on the array but is absent on the host usually reflects a missing rescan or a masking/host-group mismatch. Rescanning the bus and confirming the initiator is registered in the correct host group addresses both causes without destructive action.

Question 14Storage (Expert-level track content, domains not enumerated in source)

Which feature enables a storage array to continue serving I/O when an entire controller node fails in a dual-controller system?

  1. Quality of service throttling
  2. Thin provisioning
  3. Controller failover with LUN ownership transfer Correct answer.
  4. Snapshot scheduling

Explanation

In a dual-controller array, each LUN has an owning controller; on node failure the surviving controller takes over ownership and re-presents the LUNs, preserving access. Snapshot scheduling, thin provisioning, and QoS are data services that do not provide node-level availability.

Question 15Storage (Expert-level track content, domains not enumerated in source)

What is the effect of enabling deduplication on a backup target that stores many full backups of similar virtual machines?

  1. It increases the RPO of replication
  2. It eliminates the need for backup verification
  3. It reduces physical capacity by storing repeated blocks once Correct answer.
  4. It converts full backups into incremental backups

Explanation

Deduplication identifies repeated block content across backups and stores a single instance with references, greatly reducing physical capacity for similar full backups. It does not change replication RPO, remove verification duties, or alter backup types.

Question 16Storage (Expert-level track content, domains not enumerated in source)

Which statement best explains why an all-flash array can achieve lower latency than a hybrid array for random reads?

  1. Hybrid arrays cannot use cache
  2. SSDs always use larger block sizes
  3. SSDs have no seek time and provide uniform low access latency Correct answer.
  4. All-flash arrays do not require RAID

Explanation

Flash media have no mechanical positioning delay, so random reads incur consistent low latency rather than the seek and rotational delays of HDDs. Hybrid arrays do use cache, all-flash arrays still require RAID protection, and block size is a configuration choice.

Question 17Storage (Expert-level track content, domains not enumerated in source)

A consistency group spans several LUNs used by a clustered application. During a planned migration, in what order should the group be handled?

  1. Suspend or split the entire group together so all members switch at the same point Correct answer.
  2. Split members one at a time in any order
  3. Delete the group before migration
  4. Migrate only the largest LUN first

Explanation

Members of a consistency group must be switched together so the application sees a single consistent point in time across all LUNs. Handling members individually risks cross-LUN inconsistency, and deleting the group removes the coordination mechanism.

Question 18Storage (Expert-level track content, domains not enumerated in source)

Which QoS mechanism is most appropriate to prevent a noisy-neighbor LUN from saturating array front-end ports?

  1. RAID level change
  2. Initiator CHAP authentication
  3. Per-LUN IOPS and bandwidth upper limits Correct answer.
  4. Snapshot reserve sizing

Explanation

Upper-limit QoS policies cap IOPS and bandwidth for a specific LUN or host group, containing the impact of a heavy consumer on shared front-end resources. Snapshot reserve, RAID level, and CHAP authentication do not regulate I/O rates.

Question 19Storage (Expert-level track content, domains not enumerated in source)

What does a storage array do during a RAID rebuild after a disk failure?

  1. It restores the failed disk from the last full backup
  2. It suspends all host I/O until the rebuild finishes
  3. It converts the RAID group to RAID 0
  4. It reconstructs the failed member's data from parity or mirror data onto a spare Correct answer.

Explanation

Rebuild reads surviving members and recomputes the lost data from parity or mirror copies, writing it to a hot spare or replacement disk. It does not use backups, halt all host I/O, or downgrade the RAID level.

Question 20Storage (Expert-level track content, domains not enumerated in source)

A company needs to prove that its replicated data at a DR site is usable without disrupting production. Which approach is appropriate?

  1. Perform a test failover on an isolated network using a snapshot of the replica Correct answer.
  2. Rely on the replication link status alone
  3. Delete the replication pair to verify the data
  4. Run a full backup at the production site only

Explanation

A test failover uses a snapshot or clone of the replica on an isolated network so the DR copy can be validated without affecting production or the replication relationship. Deleting the pair is destructive, production backups do not validate the replica, and link status does not prove data usability.

Question 21Storage (Expert-level track content, domains not enumerated in source)

Which factor most limits the distance over which synchronous replication can be deployed?

  1. The size of the snapshot reserve
  2. Round-trip latency between sites and its impact on host write response time Correct answer.
  3. The RAID level of the primary pool
  4. The number of LUNs in the consistency group

Explanation

Synchronous replication waits for secondary acknowledgement before completing the host write, so round-trip latency directly adds to host write response time and constrains distance. LUN count, RAID level, and snapshot reserve do not set the distance limit.

Question 22Storage (Expert-level track content, domains not enumerated in source)

What is the purpose of host LUN masking on a storage array?

  1. To encrypt LUN data at rest
  2. To restrict which initiators can access a given LUN Correct answer.
  3. To compress LUN data
  4. To tier LUN data across media

Explanation

LUN masking maps a LUN only to specified host initiators or host groups, preventing unauthorized hosts from seeing or writing to it. Encryption, compression, and tiering are separate data services.

Question 23Storage (Expert-level track content, domains not enumerated in source)

An administrator observes that a thin-provisioned pool is approaching its subscription limit while actual used capacity is much lower. What is the most appropriate action?

  1. Convert all LUNs to snapshots
  2. Disable deduplication immediately
  3. Ignore it because thin provisioning never exhausts
  4. Monitor and expand the pool or reclaim space before allocation failures occur Correct answer.

Explanation

Thin provisioning allows subscription to exceed physical capacity, so the pool must be monitored and expanded, or space reclaimed, before allocation requests fail. Thin provisioning can indeed exhaust, snapshots do not solve capacity, and disabling deduplication usually increases consumption.

Question 24Storage (Expert-level track content, domains not enumerated in source)

Which combination best describes how an all-flash array maintains endurance while delivering high write throughput?

  1. Increasing read-ahead depth
  2. Wear leveling, write amplification reduction, and spare capacity provisioning Correct answer.
  3. Disabling RAID and using single disks
  4. Larger snapshot reserves and more frequent backups

Explanation

Flash endurance is preserved by distributing writes evenly across cells, reducing write amplification through techniques such as inline deduplication and compression, and keeping spare capacity for garbage collection. Snapshots, disabling RAID, and read-ahead do not address media wear.

Question 25Storage (Expert-level track content, domains not enumerated in source)

What is the function of a hot spare disk in a RAID group?

  1. It is an idle disk available to automatically replace a failed member and trigger rebuild Correct answer.
  2. It holds replication logs
  3. It caches host writes
  4. It stores snapshot copies

Explanation

A hot spare is an unassigned disk reserved so the array can automatically substitute it for a failed member and begin reconstruction, restoring redundancy quickly. It does not store snapshots, cache writes, or hold replication logs.

Question 26Storage (Expert-level track content, domains not enumerated in source)

In a Huawei OceanStor multi-controller architecture, what is the primary role of the SVP (Service Processor) management module?

  1. It acts as the metadata master that arbitrates LUN ownership across controllers
  2. It provides out-of-band management and fault monitoring independent of the data path Correct answer.
  3. It carries production host I/O between the front-end ports and the disk enclosures
  4. It performs inline deduplication and compression for all written blocks

Explanation

The SVP is a management and monitoring subsystem that runs out-of-band, so administrators can log in, collect logs, and receive alarms even when the data path is impaired. It does not move host I/O and it does not own storage metadata, which belong to the controllers and their clustered software.

Question 27Storage (Expert-level track content, domains not enumerated in source)

An OceanStor array uses RAID 6 for a 12-disk SAS group. Which statement correctly describes the fault tolerance of this RAID level?

  1. It survives any three disk failures because of distributed parity
  2. It survives the failure of one disk and uses the spare for the second
  3. It survives two failures only after a hot spare has been rebuilt into the group
  4. It survives the simultaneous failure of any two member disks Correct answer.

Explanation

RAID 6 uses two independent parity blocks per stripe, so the array can reconstruct data after any two member disks fail at the same time. A hot spare participates in rebuild but is not part of the fault-tolerance definition of the RAID level.

Question 28Storage (Expert-level track content, domains not enumerated in source)

Which statement describes the purpose of the OceanStor DeviceManager alarm and event log?

  1. It holds the host multipath configuration for failover
  2. It records component faults and state changes so administrators can audit and troubleshoot the array Correct answer.
  3. It stores the write cache contents until they are destaged to disk
  4. It archives the array configuration database for controller replacement

Explanation

Alarm and event logs capture hardware and software events with timestamps and severity so faults can be traced and audited. They are diagnostic records, not a cache, a multipath store, or the configuration backup itself.

Question 29Storage (Expert-level track content, domains not enumerated in source)

What does an iSCSI target portal group (TPG) control on a Huawei storage array?

  1. Which target IP addresses and ports a host may use to reach a specific iSCSI target Correct answer.
  2. Which TCP window size the array negotiates
  3. Which LUNs are mapped to the host initiator
  4. Which CHAP secret the host must present

Explanation

A target portal group binds a set of target IP addresses and listening ports to an iSCSI target, defining the access paths a host can use. LUN-to-host mapping is controlled by LUN masking, and CHAP is a separate authentication setting.

Question 30Storage (Expert-level track content, domains not enumerated in source)

Which capability distinguishes a HyperMetro active-active stretched cluster from a conventional active-passive remote replication setup?

  1. Both sites can serve host I/O concurrently for the same LUN with cross-site consistency Correct answer.
  2. The secondary site is mounted only after an administrator promotes it manually
  3. Only the primary site accepts writes while the secondary site holds a read-only copy
  4. Replication is always asynchronous to avoid dependence on link latency

Explanation

HyperMetro presents the same LUN from both sites at once, so hosts can read and write at either site while the array keeps the two copies consistent. Conventional replication keeps the secondary copy passive until a failover is declared.

Question 31Storage (Expert-level track content, domains not enumerated in source)

Which statement correctly describes the function of a LUN copy (clone) created from a snapshot in Huawei storage?

  1. It always consumes zero capacity in the storage pool
  2. It keeps sharing source blocks until the source is deleted
  3. It creates an independent full copy whose data diverges from the source after creation Correct answer.
  4. It can only be read by the host that owns the source LUN

Explanation

A clone materializes a separate LUN that is initially populated from the source, and subsequent writes to either LUN do not affect the other. This independence is what distinguishes a clone from a snapshot, which continues to reference source data.

Question 32Storage (Expert-level track content, domains not enumerated in source)

Which protocol does OceanStor use to provide block-level access over a converged Ethernet fabric while preserving lossless delivery?

  1. NFSv4.1 over a standard routed IP network
  2. SMB 3.0 over TCP with multichannel enabled
  3. FCoE with priority flow control and DCB configured end to end Correct answer.
  4. iSCSI over a congested best-effort VLAN

Explanation

FCoE carries Fibre Channel frames over Ethernet and relies on Data Center Bridging, including priority flow control, to make the link lossless. NFS and SMB are file protocols, and plain iSCSI over a best-effort VLAN does not provide lossless delivery.

Question 33Storage (Expert-level track content, domains not enumerated in source)

What is the effect of setting a LUN's write cache policy to write-through instead of write-back?

  1. Writes bypass the storage pool entirely
  2. Writes are deduplicated before being acknowledged
  3. Writes are acknowledged as soon as they reach controller memory
  4. Writes are acknowledged only after being committed to the back-end disks Correct answer.

Explanation

Write-through forces each write to be persisted on the back-end media before the host receives acknowledgment, trading latency for stronger durability. Write-back, by contrast, acknowledges from protected cache and destages later.

Question 34Storage (Expert-level track content, domains not enumerated in source)

A host writes 64 KB random I/O to a thick LUN on a hybrid array. Which array-side factor most directly determines the read latency observed by the host?

  1. The size of the LUN's metadata region in the pool
  2. The number of host HBA queue depth entries configured
  3. The frequency of snapshot schedules on unrelated LUNs
  4. The probability that the requested blocks are already resident in SSD-tier cache Correct answer.

Explanation

On a hybrid array, random reads are fast only when the blocks have been promoted to the SSD tier; a miss falls through to HDD and adds mechanical latency. HBA queue depth affects concurrency rather than the service time of a given read.

Question 35Storage (Expert-level track content, domains not enumerated in source)

An administrator must protect a production LUN with a recovery point objective of zero and a recovery time objective of minutes across two data centers 40 km apart. Which design best meets both targets?

  1. Backup to tape with weekly offsite rotation
  2. Local snapshot scheduling every 30 minutes
  3. Asynchronous remote replication with a 15-minute interval
  4. HyperMetro active-active synchronous mirroring with automatic failover Correct answer.

Explanation

An RPO of zero requires every write to be committed at both sites before acknowledgment, which synchronous active-active mirroring provides, and automatic failover meets the minutes-level RTO. Asynchronous replication, snapshots, and tape all lose some recent writes.

Question 36Storage (Expert-level track content, domains not enumerated in source)

A storage pool shows 70 percent raw capacity used but hosts report that mapped thin LUNs are nearly full. Which action addresses the immediate risk of pool exhaustion?

  1. Increase the host file system block size
  2. Expand the pool by adding disks or migrate LUNs to a larger pool Correct answer.
  3. Enable compression on the pool and wait for background reclaim
  4. Reduce the LUN snapshot retention period

Explanation

When raw pool consumption approaches its limit, the durable fix is to add capacity or relocate data so that incoming writes have somewhere to land. Compression and snapshot cleanup may help over time but cannot be relied on to avert imminent exhaustion.

Question 37Storage (Expert-level track content, domains not enumerated in source)

A clustered application uses several LUNs in a consistency group replicated to a DR site. What property does the consistency group guarantee at the secondary site?

  1. Each LUN is recovered independently at its own latest point
  2. Only the first LUN in the group is replicated
  3. All member LUNs are recovered to the same point in time Correct answer.
  4. Member LUNs are recovered in the order they were created

Explanation

A consistency group coordinates replication so that a recovery uses a single common point in time across all member LUNs, preserving cross-LUN write ordering for the application. Independent per-LUN recovery points could leave the cluster with inconsistent data.

Question 38Storage (Expert-level track content, domains not enumerated in source)

An engineer enables SmartQoS on a LUN used by a batch job that competes with an online transaction system. What is the intended outcome of assigning an I/O priority and upper limit to the batch LUN?

  1. The batch LUN is converted from thick to thin provisioning
  2. Batch I/O is throttled so the transaction LUN retains its latency target Correct answer.
  3. Batch I/O is redirected to a dedicated physical disk group
  4. The transaction LUN is automatically moved to the SSD tier

Explanation

SmartQoS applies per-LUN or per-host I/O limits and priorities so a low-priority workload cannot consume more than its share of array resources. It shapes traffic rather than relocating data or changing provisioning type.

Question 39Storage (Expert-level track content, domains not enumerated in source)

What is the primary purpose of configuring CHAP authentication on an iSCSI connection between a host and an OceanStor array?

  1. To encrypt data blocks in transit between host and array
  2. To verify the identity of the initiator before granting access to the target Correct answer.
  3. To discover target portals automatically
  4. To compress iSCSI PDUs on the wire

Explanation

CHAP is a challenge-response authentication mechanism that proves the initiator's identity to the target, preventing unauthorized hosts from logging in. It provides authentication only; it does not encrypt, compress, or discover portals.

Question 40Storage (Expert-level track content, domains not enumerated in source)

A customer wants to add a second storage pool to an existing OceanStor array with no downtime. Which sequence is correct?

  1. Convert the existing LUNs to thin provisioning before adding disks
  2. Delete the existing pool first to free disks for the new pool
  3. Power off the array, install disks, and rebuild the existing pool
  4. Create the pool on available disks, then create and map LUNs and rescan on the host Correct answer.

Explanation

Pool creation is an online operation: the array uses free disks to build the new pool, after which LUNs can be created and mapped and the host rescans its bus. Deleting the existing pool or powering off the array would cause unnecessary disruption.

Question 41Storage (Expert-level track content, domains not enumerated in source)

Which statement explains why erasure coding is often preferred over 3-way replication for capacity efficiency in a distributed storage system?

  1. Erasure coding requires fewer nodes than replication
  2. Erasure coding replicates every block three times across nodes
  3. Erasure coding stores parity fragments so usable capacity exceeds one third of raw Correct answer.
  4. Erasure coding eliminates the need for any redundancy

Explanation

Erasure coding splits data into fragments with parity, so overhead is a fraction of the data rather than two full extra copies. Three-way replication uses three times the raw capacity to protect the same data.

Question 42Storage (Expert-level track content, domains not enumerated in source)

An administrator observes that a host sees duplicate paths to the same LUN after zoning changes. Which action resolves the duplicate presentation while preserving redundancy?

  1. Configure multipathing so the extra paths are grouped as standby or active-active Correct answer.
  2. Remove all but one path to the LUN
  3. Delete and recreate the host mapping
  4. Disable the array's front-end ports on one controller

Explanation

Multipathing software aggregates multiple physical paths to a single LUN into one logical device, using the extra paths for load balancing or failover. Removing paths or ports would eliminate the redundancy the design intends to keep.

Question 43Storage (Expert-level track content, domains not enumerated in source)

Which behavior is expected when a snapshot's reserved space in a thin pool is exhausted?

  1. The snapshot automatically converts into a full clone
  2. The snapshot becomes invalid and the source LUN may be affected by pool pressure Correct answer.
  3. The source LUN is unmapped from all hosts
  4. The array silently deletes the oldest snapshot

Explanation

Snapshots rely on copy-on-write space to preserve pre-modification blocks; if that space runs out, the snapshot can no longer be maintained and the pool may come under pressure. The array does not silently convert the snapshot to a clone or delete snapshots without policy.

Question 44Storage (Expert-level track content, domains not enumerated in source)

A NAS share on an OceanStor system must be accessible to both Windows and Linux clients with identity mapping. Which configuration is required?

  1. An iSCSI LUN formatted with NTFS and shared by one server
  2. A CIFS share with an AD domain joined and LDAP or NIS mapping for UNIX identities Correct answer.
  3. An FCoE zone containing both client HBAs
  4. An NFS export with anonymous UID mapping only

Explanation

Multi-protocol access requires the array to resolve both Windows and UNIX identities, typically by joining Active Directory and mapping UNIX users through LDAP or NIS. Anonymous-only NFS or block protocols cannot provide the shared identity model needed.

Question 45Storage (Expert-level track content, domains not enumerated in source)

What is the effect of enabling read-only cache (cache read-ahead) on a LUN serving sequential media streaming?

  1. The array compresses data before writing it to disk
  2. The array duplicates writes to both controllers for protection
  3. The array pins all read data permanently in cache
  4. The array prefetches upcoming blocks to reduce read latency for sequential access Correct answer.

Explanation

Read-ahead detects sequential patterns and prefetches following blocks into cache so subsequent reads are served from memory. It does not duplicate writes, compress data, or pin data permanently.

Question 46Storage (Expert-level track content, domains not enumerated in source)

A storage administrator must verify that a newly configured HyperMetro pair will not introduce write latency above the application's tolerance. Which measurement is most relevant?

  1. The host operating system's page cache size
  2. Round-trip latency between the two sites' array front-end ports Correct answer.
  3. The RAID level of the underlying disk groups
  4. The number of LUNs in the pair's consistency group

Explanation

Synchronous mirroring acknowledges a write only after both sites commit it, so inter-site round-trip latency directly adds to host write latency. RAID level and consistency group size affect other characteristics but not the propagation delay.

Question 47Storage (Expert-level track content, domains not enumerated in source)

During a controller replacement on a dual-controller OceanStor array, which step ensures the new controller adopts the correct identity and configuration?

  1. Import the previously exported configuration file and verify serial numbers before service Correct answer.
  2. Recreate all LUNs manually from the host side
  3. Disable the surviving controller to force a clean takeover
  4. Reset both controllers to factory defaults and re-zone the fabric

Explanation

Configuration backup and restore preserves LUN mappings, pools, and system identity so the replacement controller joins the cluster consistently. Manual recreation or factory reset risks data loss and misconfiguration.

Question 48Storage (Expert-level track content, domains not enumerated in source)

A customer reports intermittent path loss on an iSCSI LUN during peak hours while the array shows no port errors. Which investigation is most likely to reveal the cause?

  1. Review the array's snapshot retention policy
  2. Examine switch port utilization, pause frames, and host NIC error counters for congestion Correct answer.
  3. Verify the LUN's thin provisioning subscription rate
  4. Check the LUN's RAID level and rebuild status

Explanation

Intermittent loss without array port errors points to network congestion or frame drops along the path, so switch utilization, flow control, and host NIC counters are the right evidence. RAID, snapshot, and provisioning settings do not explain transient connectivity loss.

Question 49Storage (Expert-level track content, domains not enumerated in source)

An engineer must design a backup target that stores many similar full backups while keeping restore times acceptable. Which combination is most appropriate?

  1. Snapshot-only protection with no separate backup copy
  2. Triple replication with no deduplication
  3. RAID 0 across all disks for maximum write speed
  4. Deduplication with post-process or inline processing plus sufficient cache for restore reads Correct answer.

Explanation

Deduplication removes repeated blocks across similar full backups, while adequate cache and SSD tiering keep restore reads fast. RAID 0 offers no protection and snapshot-only protection does not provide an independent backup copy.

Question 50Storage (Expert-level track content, domains not enumerated in source)

Which statement best explains why an all-flash array can sustain low latency under mixed read/write load compared with a hybrid array?

  1. Writes are absorbed by SSD-backed cache and destaged without HDD seek penalties Correct answer.
  2. The array disables garbage collection to avoid write amplification
  3. The array uses larger stripe sizes to reduce parity computation
  4. Reads are served from HDD while writes go to SSD

Explanation

In an all-flash array both reads and writes land on media without mechanical seek time, and write cache absorbs bursts before destaging to SSD. Hybrid arrays still incur HDD latency when cache misses or destaging occurs.