Storage practice questions

Worked examples for Storage

1. Question 1

   During a design review, a colleague proposes storing a shared media library that must be read concurrently by dozens of instances, and a separate high-IOPS transactional database, both on the same set of block volumes. Which principle should guide the architect's response about Block Volume's role among OCI storage services?

   1. A. Block Volume is the correct choice for both, since a single block volume can be mounted read-write by an unlimited number of instances without additional software.

      Overstates block sharing. Attaching a volume to multiple instances requires cluster-aware handling and is not a substitute for a shared file system; unlimited read-write mounting is not a Block Volume behavior.

   2. B. Object Storage should host both workloads, because it exposes block devices to instances at lower cost than Block Volume.

      Misdescribes Object Storage. It stores objects accessed through an API rather than presenting block devices, so it cannot back a transactional database's block I/O.

   3. C. Block Volume provides raw block storage attached to instances and is the right fit for the database, while broadly shared, concurrently accessed content is better served by a shared file or object storage service.Correct answer

      Block Volume presents unformatted block devices to individual instances, which suits databases and file systems, whereas widely shared or internet-accessible content belongs on File Storage or Object Storage.

   4. D. The decision depends only on the elastic performance level chosen, since a high enough level lets a block volume behave as a shared file system.

      Confuses performance tiering with access semantics. The performance level changes IOPS and throughput; it does not give a block device shared file-system semantics.

   Explanation

   Block Volume provides persistent, unformatted block devices attached to compute instances over the network, which is the right substrate for databases and instance file systems needing block-level I/O. Content that many clients must read concurrently fits a shared file service or Object Storage, which is designed for durable, API-accessible objects rather than block devices. Attaching one block volume to many instances read-write demands cluster-aware software and is not equivalent to a shared file system, and no elastic performance level changes a block device's access semantics.

2. Question 2

   An architect is documenting first-byte latency expectations so application teams can pick a tier safely. Which statement correctly contrasts the retrieval behaviour of the Infrequent Access tier with that of the Archive tier?

   1. A. Infrequent Access objects are read directly with low first-byte latency comparable to Standard, while Archive objects are unreadable until a restore completes, giving a first-byte time measured in a much longer intervalCorrect answer

      Correct. Infrequent Access trades a per-GB retrieval fee for immediate, Standard-like read performance, whereas Archive adds an explicit restore step that dominates first-byte time (Object Storage tiers).

   2. B. Both tiers require a restore request; Infrequent Access simply completes its restore faster than Archive

      Treats Infrequent Access as a fast Archive. Infrequent Access has no restore step at all — it is directly readable — so describing it as a quicker restore misstates the tier.

   3. C. Neither tier requires a restore; the only difference between them is the per-GB storage price

      Erases Archive's restore requirement, which is the single most important operational difference between the two tiers and the reason Archive cannot serve interactive reads.

   4. D. Infrequent Access reads are slower than Archive reads because Infrequent Access data is stored on lower-performance media

      Invents a media-performance story and inverts the ordering. Infrequent Access serves reads with the same immediacy as Standard; the price difference reflects the retrieval fee model, not slower hardware.

   Explanation

   Infrequent Access and Archive both reduce storage cost, but only Archive changes the retrieval semantics: an Archive object must be restored before it can be read, so its effective first-byte time is dominated by the restore rather than by network latency. Infrequent Access data is read directly with latency comparable to Standard and pays for that with a per-GB retrieval fee. Any description that gives Infrequent Access a restore step, that denies Archive's restore step, or that claims Infrequent Access is the slower of the two misrepresents the tiers and would lead teams to break interactive access SLAs.

3. Question 3

   A development team needs a writable, fully independent copy of a 2 TB production block volume in the same region so they can attach it to a test instance and begin work within minutes. The copy has no retention or archival requirement — it will be deleted after the test cycle. Which Block Volume capability best matches this requirement?

   1. A. Create a clone of the production volume; the clone is an independent volume in the same region that can be attached while the copy completes in the background.Correct answer

      Correct. A clone is a point-in-time, disk-to-disk copy created within the same region that becomes an independent volume usable almost immediately, which is exactly the fast-working-copy use case.

   2. B. Create a manual full backup of the production volume, then attach that backup directly to the test instance.

      Assumes a backup is an attachable object. A block volume backup is a point-in-time copy held in Object Storage; it must first be restored into a new block volume before any instance can attach it.

   3. C. Assign a scheduled backup policy to the production volume so the test team can mount the most recent scheduled backup.

      Confuses retention scheduling with provisioning a working copy. Backup policies automate point-in-time copies for retention and recovery; they still produce backups that require a restore, and scheduling adds latency the team does not need.

   4. D. Copy a backup of the production volume to a second region and restore it there, then attach the restored volume over the region interconnect.

      Applies a cross-region DR mechanism to an in-region provisioning need. Cross-region backup copy exists to survive a regional event; here it only adds copy and restore time, and a volume can only be attached to an instance in its own region.

   Explanation

   Clones and backups solve different problems. A clone is an immediate, independent disk-to-disk copy of a volume within the same region that can be attached and used right away, which fits ephemeral test or development environments. Backups are point-in-time copies stored in Object Storage for retention and recovery; they are not attachable and must be restored into a new volume first, and scheduling them through a policy or copying them to another region only adds time to an in-region provisioning task.

4. Question 4

   A compliance policy requires that an application's Block Volume data remain recoverable even if the entire OCI region hosting it becomes unavailable, and that the recovery procedure be exercised in the secondary region. Which design meets the requirement using Block Volume backup capabilities?

   1. A. Schedule backups with a backup policy and copy the backups to a second region, then restore a backup into a new volume there when recovery is needed.Correct answer

      Backups are stored in Object Storage and can be copied to another region; recovery works by restoring a copied backup into a new volume in that region, which is the supported cross-region DR path (Block Volume overview).

   2. B. Clone the volume, since a clone is automatically placed in a second region for disaster-recovery purposes.

      Misstates cloning as a cross-region mechanism; a clone is created within the region as an immediate working copy and provides no protection against loss of that region.

   3. C. Rely on a higher elastic performance level, because higher-performance volumes are replicated across regions as part of that level.

      Ties durability and geographic redundancy to the performance level, which only governs IOPS and throughput and has nothing to do with where copies are kept.

   4. D. Attach the backup directly to a compute instance in the second region, since a copied backup can be mounted without being restored.

      Repeats the misconception that a backup is directly attachable; a backup must first be restored into a new volume before any instance can attach it.

   Explanation

   Block Volume backups are point-in-time copies kept in Object Storage, and they can be copied to another region so the data survives the loss of the original region; recovery then means restoring a copied backup into a new volume in that region and attaching it. Cloning is an in-region operation that produces an immediately usable duplicate and offers no regional protection, and a backup is never mountable in place — the restore step is what yields an attachable volume. The elastic performance level governs only IOPS and throughput and carries no cross-region replication semantics.

5. Question 5

   A team has configured cross-region replication from a production bucket to a bucket in a second region. They describe the setup to their auditor as protection against a user accidentally deleting production data. Why is that claim incorrect, and what should be added?

   1. A. Deletions in the source bucket propagate to the destination, so the copy disappears too; versioning on the bucket, or retention rules, is what protects against accidental deletionCorrect answer

      Correct. Replication is a mirror rather than a backup — it applies source deletions to the destination — so accidental-deletion protection comes from versioning or retention rules (Using Replication; Using Object Versioning).

   2. B. The claim is correct as written, because replication keeps an independent copy that deletions in the source never touch

      Treats replication as a backup. Replication mirrors the source's state, deletions included, so the destination does not preserve data the source no longer has.

   3. C. Replication only copies objects once at policy creation, so anything written afterwards is unprotected; scheduling a second replication policy fixes it

      Wrong: replication is continuous for objects written after the policy is created, not a one-time copy, so "schedule a second policy" is unnecessary. (Replication also does not backfill objects that existed before the policy.) Either way, ongoing replication propagates source deletes to the destination, so it is not protection against accidental deletion.

   4. D. Replication skips objects in colder tiers, so the fix is to keep everything in Standard

      Substitutes an unrelated tiering concern for the real gap. The reason replication is not deletion protection is that deletes propagate, not that some tiers are excluded.

   Explanation

   Object Storage replication asynchronously keeps a destination bucket in step with a source bucket, and that includes applying deletions — which makes it a resilience and locality mechanism, not a defence against human error or malicious deletion. Guarding against accidental deletion requires versioning, which retains prior versions when an object is deleted or overwritten, or retention rules, which block deletion outright for a period. Replication is not a single point-in-time copy, and its limitation here has nothing to do with which storage tier the objects occupy.

6. Question 6

   A database instance stores its data files, redo logs, and archive logs on three separate block volumes attached to the same compute instance. The architect must be able to recover all three volumes to the same instant so the database can be recovered consistently. Which OCI Block Volume construct is designed for this requirement?

   1. A. Assign the same backup policy to each of the three volumes individually so they all follow an identical schedule.

      Confuses an identical schedule with a coordinated point in time. Three independently scheduled backups start as separate operations, so their point-in-time snapshots can differ and the recovered set may be mutually inconsistent.

   2. B. Place the three volumes in a volume group and back up the volume group, which produces a crash-consistent point-in-time backup across all member volumes.Correct answer

      Correct. A volume group is exactly the construct for multi-volume coordination: a volume group backup (or clone) captures every member volume at the same point in time, giving crash-consistent recovery for multi-disk workloads.

   3. C. Create a clone of each volume in sequence, relying on clone creation being fast enough that the three copies are effectively simultaneous.

      Substitutes speed for a consistency guarantee. Sequential clone operations have independent point-in-time markers; 'close enough in time' is not crash consistency, and writes between operations can leave the set out of step.

   4. D. Enable cross-region backup copy on each volume, since the copy operation synchronizes the three volumes into a single consistent set in the destination region.

      Attributes a consistency function to a replication feature. Cross-region copy moves an already-taken backup to another region; it does not coordinate the point in time at which those backups were captured.

   Explanation

   Recovering a multi-volume workload consistently requires all member volumes to be captured at the same instant, and the volume group is the construct that provides that: a volume group backup or volume group clone is crash-consistent across every volume in the group. Independent per-volume backups — even on identical schedules — and sequential clones each carry their own point in time, so the recovered set can be internally inconsistent. Cross-region backup copy addresses where a copy lives, not whether multiple volumes share a consistent point in time.

7. Question 7

   A team stores large media archives that are read only a few times per year, mostly as long sequential scans, on block volumes attached to an analysis instance. They want to reduce cost without moving off Block Volume. What is the appropriate design choice?

   1. A. Keep the volumes at a high-performance level, because lowering the level of an existing volume would require destroying and recreating it.

      Represents the misconception that the performance level is immutable. The elastic performance level of an existing volume can be changed in either direction.

   2. B. Assign the volumes the lower-cost, lower-IOPS elastic performance level, which targets infrequently accessed and sequentially read data.Correct answer

      The lower-cost elastic performance level is intended for workloads with modest IOPS needs such as cold, large, sequentially accessed data, so it reduces cost while still meeting the access pattern.

   3. C. Delete the volumes and keep only manual backups, attaching a backup whenever an analyst needs to read the archive.

      Assumes a backup is directly attachable. Reading from a backup requires restoring it into a new volume first, so this is not a working access pattern for periodic reads.

   4. D. Convert the volumes into a volume group, since group members are billed at a single discounted rate.

      Invents a pricing behavior for volume groups. A volume group is a management construct for coordinated backups and clones and does not change how member volumes are charged.

   Explanation

   Elastic performance levels let each volume's IOPS and throughput characteristics be matched to its workload, and the lower-cost level is aimed at data that is accessed infrequently or read sequentially — exactly the media-archive pattern described. Because the level is elastic, there is no need to preserve a high-performance setting out of fear of recreation, since an existing volume's level can be lowered. Replacing volumes with backups is not a usable read path because a backup must be restored before it can be attached, and a volume group changes management semantics rather than billing.

8. Question 8

   An administrator has defined and enabled a lifecycle policy on a bucket, but no objects are ever transitioned or deleted even though many clearly satisfy the rule conditions. Which prerequisite is most likely missing?

   1. A. Versioning must be enabled on the bucket before any lifecycle rule can execute

      Believes lifecycle depends on versioning. The two features are independent and can coexist; versioning only adds the option of targeting previous object versions in a rule.

   2. B. The bucket must be in the Archive tier for archive actions to be permitted

      Inverts the purpose of an archive action. The rule is what moves objects into Archive; requiring the bucket to already be Archive would make the action pointless.

   3. C. A replication policy must exist so the service has a target to write the transitioned objects to

      Confuses lifecycle with replication. Tier transitions happen in place within the same bucket and have no dependency on a replication target.

   4. D. An IAM policy granting the Object Storage service permission to manage objects in the compartment or tenancy has not been createdCorrect answer

      Correct. Lifecycle actions are executed by the Object Storage service principal, which needs an explicit policy such as allowing the objectstorage service to manage object-family; without it, rules are defined but take no effect (Using Object Lifecycle Management — required policy).

   Explanation

   Lifecycle actions are carried out by the Object Storage service itself rather than by the administrator's own credentials, so the tenancy must contain an IAM policy authorising the Object Storage service principal to manage the object family in the relevant compartment. Without that grant the rules exist and appear enabled but nothing is archived or deleted. Versioning is an independent bucket feature and is not a precondition, an archive action is what places objects into the Archive tier rather than something requiring an Archive bucket, and replication is a separate policy that plays no part in in-place tier transitions.