Top 30 Storage Interview Questions and Answers

30 questions. Pick an answer, or open “Show the answer” to read it.

Answers are graded in your browser. Nothing is saved, and no XP or streak is earned here. The app keeps score.

1. Question 1 of 30

   Which statement best captures the mechanical cost of maintaining multiple indexes on a write-heavy table?

   • aEach write requires updating only the indexes whose key columns changed, minimizing I/O for unrelated indexes
   • bThe database uses binary search trees for indexes, causing frequent full-tree rebalancing on every insert
   • cEvery table write typically triggers random I/O to update each index's B-Tree, plus node splits and log overhead
   • dIndexes are stored as hash maps, so range queries require full table scans but writes remain cheap
   Show the answer

   Answer: c · Every table write typically triggers random I/O to update each index's B-Tree, plus node splits and log overhead

   The card explains that every write likely updates every index, causing extra random I/O, node splits, and WAL overhead. Option B reflects the common misconception that binary search trees are the classic disk structure, while D confuses hash indexes with the standard B-Tree approach.

   Read the full bite: Explain database indexes, the classic data structure, and write-heavy trade-offs

2. Question 2 of 30

   Which statement correctly describes the relationship between OCI image layers and OverlayFS in container runtime?

   • aOCI stores images as monolithic blobs that OverlayFS splits into lowerdir and upperdir at runtime
   • bOCI specifies layer tarballs and manifests, while OverlayFS is the in-kernel driver that assembles them at runtime
   • cOverlayFS defines the image format and compression standard used by container registries
   • dOCI layers are unpacked and merged by OverlayFS in userland before the container starts
   Show the answer

   Answer: b · OCI specifies layer tarballs and manifests, while OverlayFS is the in-kernel driver that assembles them at runtime

   OCI governs the packaging and distribution of images as tarballs and manifests, while OverlayFS is solely a Linux kernel filesystem driver that mounts those layers at runtime. Distractor B is a common misconception that conflates the runtime driver with the image specification itself.

   Read the full bite: Explain layered filesystems like OverlayFS and their efficiency vs monolithic models

3. Question 3 of 30

   What is the primary reason Union File Systems, like Docker's OverlayFS, can exhibit performance overhead for write-intensive applications?

   • aEach modification to an existing file from a lower layer triggers a copy-on-write operation.
   • bAll write operations are first buffered in memory before being committed to disk, causing delays.
   • cThey require frequent re-merging of all layers for every write operation.
   • dThe system must synchronize changes across all containers sharing the same base image.
   Show the answer

   Answer: a · Each modification to an existing file from a lower layer triggers a copy-on-write operation.

   The card states that "The copy-on-write mechanism adds performance overhead for every initial write to a file that exists in a lower layer." This means files from read-only layers must first be copied to the writable layer before they can be modified, which is an extra I/O operation. Distractor A is incorrect because lower layers are read-only and not re-merged on write; changes are isolated to the top writable layer.

   Read the full bite: Union File Systems: Docker's Layered Magic

4. Question 4 of 30

   Which scenario best highlights the primary benefit of using a data lake compared to a data warehouse?

   • aFacilitating efficient, schema-enforced data loading for standardized financial analysis.
   • bStoring vast amounts of diverse, unprocessed data for future machine learning model training.
   • cEnsuring rapid, consistent performance for daily business intelligence dashboards.
   • dMaintaining a highly structured and clean dataset for regulatory compliance reporting.
   Show the answer

   Answer: b · Storing vast amounts of diverse, unprocessed data for future machine learning model training.

   The card states that data lakes are ideal for storing "massive volumes of diverse, unstructured data cheaply, without having to define its structure or purpose upfront" and for "data scientists and machine learning engineers who need access to raw data for exploratory analysis and model training." Option B directly reflects this core advantage. Option C describes a use case better suited for a data warehouse, which is optimized for consistent, high-performance reporting.

   Read the full bite: Data Lake vs. Data Warehouse: Raw vs. Refined Data

5. Question 5 of 30

   Which scenario is the most appropriate use case for sessionStorage?

   • aTemporarily saving progress on a multi-page form within a single browser tab.
   • bStoring a user's dark mode preference to apply on every future visit.
   • cKeeping a user's authentication token secure for the duration of their session.
   • dStoring a large catalog of product information for offline browsing.
   Show the answer

   Answer: a · Temporarily saving progress on a multi-page form within a single browser tab.

   sessionStorage is ideal for temporary data tied to a single tab, such as partially filled form data that should persist across page refreshes but not after the tab closes. Storing sensitive data like authentication tokens is explicitly advised against for any Web Storage, and large datasets are better handled by IndexedDB.

   Read the full bite: Web Storage API: Browser Key-Value Stores

6. Question 6 of 30

   Which storage type is the appropriate backing for a relational database that needs low-latency random reads and writes?

   • aObject storage, because it scales to unlimited capacity
   • bBlock storage, because it provides a raw low-latency volume to a single instance
   • cObject storage, because it offers the highest durability
   • dFile storage, because it can be shared across many instances
   Show the answer

   Answer: b · Block storage, because it provides a raw low-latency volume to a single instance

   Databases need fast in-place random I/O on a volume attached to one host, which is exactly what block storage provides. Object storage reads and writes whole objects over HTTP and is unsuitable for transactional workloads.

   Read the full bite: Object vs block vs file storage

7. Question 7 of 30

   You mount a named volume at the Postgres data directory, then run docker compose down. Why does the data survive?

   • aNamed volumes are managed independently of containers, so removing containers leaves them intact
   • bdown never removes anything, only stops services
   • cPostgres replicates data to the host's RAM
   • ddown recreates the volume from a backup snapshot
   Show the answer

   Answer: a · Named volumes are managed independently of containers, so removing containers leaves them intact

   Named volumes have a lifetime separate from containers, so down (which removes containers) leaves the volume and its data; there is no snapshot, and down does remove containers unless you add -v which would also delete the volume.

   Read the full bite: Persist PostgreSQL data across compose down

8. Question 8 of 30

   Which characteristic best describes a data lake's approach to data storage and usage?

   • aIt prioritizes fast, consistent queries for standard business reporting.
   • bData is structured and validated upon ingestion to ensure immediate query readiness.
   • cIt automatically cleanses and transforms diverse datasets into a unified schema upon entry.
   • dRaw data is stored in its native format, with structure applied during analysis.
   Show the answer

   Answer: d · Raw data is stored in its native format, with structure applied during analysis.

   A data lake's defining characteristic is its "store now, structure later" approach, where raw data is kept in its native format and a schema is applied only when the data is read for analysis (schema-on-read). This provides maximum flexibility for future analysis, unlike data warehouses which impose structure upon ingestion.

   Read the full bite: Data Lake: Store Raw Data Now, Analyze It Later

9. Question 9 of 30

   In a typical web-plus-database Compose setup, which storage choice fits each service best?

   • aBind mount for both the source and the database files
   • bBind mount the source code in dev; named volume for the database data
   • cNamed volume for the source; bind mount the database files in production
   • dNamed volume for both, never use bind mounts
   Show the answer

   Answer: b · Bind mount the source code in dev; named volume for the database data

   Bind mounts suit live-reloading source in development, while named volumes give Docker-managed, portable durability ideal for database data; bind-mounting production DB files couples data to a fragile host path.

   Read the full bite: Bind mounts versus named volumes

10. Question 10 of 30

    Why is object storage generally not recommended for applications requiring frequent, small updates to data, such as a transactional database?

    • aIts flat structure makes navigating and locating specific data blocks inefficient for rapid changes.
    • bThe overhead of generating globally unique identifiers for each small update is too high.
    • cIt lacks the ability to manage file permissions and access control at a granular level.
    • dObjects are immutable, meaning even a small modification necessitates replacing the entire object.
    Show the answer

    Answer: d · Objects are immutable, meaning even a small modification necessitates replacing the entire object.

    The card explicitly states that objects are immutable, making object storage unsuitable for frequent updates because changing even one byte requires re-uploading the entire object. While other options might seem plausible, the immutability is the direct and primary reason given in the card.

    Read the full bite: Object Storage: Data Without a File Hierarchy

11. Question 11 of 30

    Why can a table have only one clustered index but many non-clustered indexes?

    • aRows can be physically ordered only one way, and the clustered index defines that order
    • bClustered indexes are larger and exhaust available storage
    • cClustered indexes do not support range queries
    • dNon-clustered indexes cannot coexist with a primary key
    Show the answer

    Answer: a · Rows can be physically ordered only one way, and the clustered index defines that order

    A clustered index sets the single physical ordering of the table's rows, so only one can exist; non-clustered indexes are separate structures with pointers, so a table can have many. Clustered indexes in fact excel at range queries, making that option wrong.

    Read the full bite: Clustered versus non-clustered indexes

12. Question 12 of 30

    Before a cloud server's operating system can store files on a newly provisioned block storage volume, what essential step must be performed?

    • aIt must be formatted with a filesystem and then mounted to a directory.
    • bIt needs to be configured as a shared network drive accessible by all virtual machines in the same subnet.
    • cThe volume must be assigned a unique public IP address for network accessibility.
    • dThe cloud provider automatically initializes it with a default file structure upon creation.
    Show the answer

    Answer: a · It must be formatted with a filesystem and then mounted to a directory.

    The card states that block storage is like a raw, unformatted SSD, and it's 'up to you to format it with a filesystem... and mount it' before use. Option B is incorrect because block storage is designed for single-server attachment, not for simultaneous access by multiple servers.

    Read the full bite: Block Storage: Your Virtual Hard Drive in the Cloud

13. Question 13 of 30

    For which scenario is cloud file storage the most suitable solution?

    • aProviding low-latency, dedicated storage for a high-performance database.
    • bSecuring individual user data on a single, isolated virtual machine.
    • cEnabling multiple servers or users to access a shared, hierarchical file system.
    • dStoring vast amounts of unstructured data for long-term archives.
    Show the answer

    Answer: c · Enabling multiple servers or users to access a shared, hierarchical file system.

    Cloud file storage is designed for scenarios requiring a shared, traditional file system accessible by multiple clients or servers, as stated in the card. Storing vast unstructured data is better suited for object storage, and high-performance databases typically use block storage.

    Read the full bite: Cloud File Storage: A Shared Drive on the Internet

14. Question 14 of 30

    A cloud storage service advertises 99.999999999% durability but only 99.9% availability. What does this primarily indicate?

    • aThe service prioritizes preventing accidental user deletions over protecting against system-wide data loss events.
    • bWhile your data is extremely unlikely to be permanently lost, it may experience periods where it cannot be accessed.
    • cYour data is guaranteed to be accessible at all times, but there's a very slight risk of corruption.
    • dData will be frequently lost due to system failures, but redundant network paths ensure constant access.
    Show the answer

    Answer: b · While your data is extremely unlikely to be permanently lost, it may experience periods where it cannot be accessed.

    High durability (11 nines) means the data is extremely unlikely to be permanently lost or corrupted. However, 99.9% availability implies that the data might be temporarily unreachable, as availability refers to reachability, not data preservation. Option C is incorrect because 99.9% availability does not guarantee accessibility at all times.

    Read the full bite: Data Durability vs. Availability: Lost vs. Unreachable

15. Question 15 of 30

    What is the primary drawback of storing data that requires frequent, low-latency access in a cold or archive storage tier?

    • aData integrity cannot be guaranteed over long periods in these tiers.
    • bThe data will be automatically deleted after a set inactivity period.
    • cAccess to the data will be restricted to administrative users only.
    • dIt will incur significant retrieval costs and access delays.
    Show the answer

    Answer: d · It will incur significant retrieval costs and access delays.

    The card states that retrieving cold data unexpectedly is slow and costly, and that retrieval fees and time delays will quickly make it more expensive than keeping it in a standard tier. Cold storage is for long-term retention, not automatic deletion.

    Read the full bite: Cloud Storage Tiers: Match Cost to Access Frequency

16. Question 16 of 30

    Which statement accurately describes the storage behavior of block storage snapshots?

    • aSubsequent snapshots only store the data blocks that have changed since the previous snapshot, referencing older data.
    • bEach snapshot creates a complete, independent copy of the volume's data at that moment.
    • cSnapshots continuously synchronize data, providing real-time data protection and consistency.
    • dDeleting an older snapshot guarantees immediate and full recovery of the storage space it occupied.
    Show the answer

    Answer: a · Subsequent snapshots only store the data blocks that have changed since the previous snapshot, referencing older data.

    The card states that subsequent snapshots only store changed data blocks and reference older data from previous snapshots, making them incremental. This also means deleting an older snapshot does not necessarily free up all its space if newer snapshots still depend on its data, making option D incorrect.

    Read the full bite: Block Storage Snapshots Are Incremental Backups

17. Question 17 of 30

    A developer configures Cross-Region Replication (CRR) for an existing S3 bucket. What is a critical behavior they should be aware of regarding the replication process?

    • aThe replication process is synchronous, ensuring immediate consistency across regions.
    • bObjects already present in the source bucket before CRR activation will not be replicated.
    • cCRR will only replicate objects that have public read access enabled.
    • dReplication only occurs for objects updated, not newly uploaded, after CRR is enabled.
    Show the answer

    Answer: b · Objects already present in the source bucket before CRR activation will not be replicated.

    The card explicitly states that CRR "does not replicate objects that were present before replication was configured." It only applies to new objects or updates made after CRR is enabled. Option A is incorrect because the card describes CRR as an "asynchronous" feature.

    Read the full bite: Cross-Region Replication (CRR): Geographic Data Copying

18. Question 18 of 30

    Which scenario highlights a key limitation of encryption at rest?

    • aData is intercepted by an attacker while being transmitted over a public network.
    • bThe encryption process introduces significant latency, making real-time data processing infeasible.
    • cData is accessed by an attacker who physically steals the storage device.
    • dAn authorized application, compromised by an attacker, reads sensitive data from storage.
    Show the answer

    Answer: d · An authorized application, compromised by an attacker, reads sensitive data from storage.

    The card explicitly states that encryption at rest "doesn't stop a compromised app with valid keys from reading" and advises "Do not rely on it to protect data from a compromised application with valid credentials." While encryption can have performance implications, the card notes that "performance overhead on modern systems is negligible."

    Read the full bite: Encryption at Rest: Securing Your Data When It's Not Moving

19. Question 19 of 30

    You are running JupyterLab in Docker and need to persist both active notebook development and multi-gigabyte datasets across container restarts. Which approach best follows Docker best practices?

    • aUse a bind mount for notebooks and a named volume for datasets, mounting the dataset read-only when appropriate
    • bAdd COPY instructions in the Dockerfile for both notebooks and datasets so they are baked into the image
    • cUse a named volume for notebooks and a bind mount for datasets to maximize portability
    • dUse bind mounts for both notebooks and datasets to keep all files on the host filesystem
    Show the answer

    Answer: a · Use a bind mount for notebooks and a named volume for datasets, mounting the dataset read-only when appropriate

    Bind mounts let host notebook edits sync immediately into the container, while named volumes persist datasets without tying them to a specific host path and remain portable across environments. Swapping them ignores the need for live code syncing, and COPY requires image rebuilds on every change.

    Read the full bite: How do you persist notebooks and artifacts in Docker?

20. Question 20 of 30

    With Object Versioning enabled, what is the immediate effect of a PUT request that would normally overwrite an existing file?

    • aA new version of the file is created, and the previous version remains accessible.
    • bThe file is temporarily locked, preventing further modifications until a new version ID is assigned.
    • cA delete marker is automatically placed on the old file, making it inaccessible.
    • dThe original file is permanently deleted and replaced by the new file.
    Show the answer

    Answer: a · A new version of the file is created, and the previous version remains accessible.

    The card explicitly states that when versioning is enabled, a PUT request 'you're not replacing it; you're creating a new version' while the old version is retained. Option D describes the behavior of storage without versioning, which is precisely what versioning aims to prevent.

    Read the full bite: Object Versioning: A Safety Net for Cloud Files

21. Question 21 of 30

    Why do databases read and write whole pages rather than individual rows directly to disk?

    • aPages prevent SQL injection at the storage layer
    • bRows are too large to write one at a time
    • cWriting rows individually would violate ACID by definition
    • dBlock-oriented I/O has a high fixed cost per operation, so whole pages amortize it across many rows
    Show the answer

    Answer: d · Block-oriented I/O has a high fixed cost per operation, so whole pages amortize it across many rows

    Each disk or SSD operation carries a large fixed cost no matter how few bytes are moved, so transferring a whole page spreads that cost over many rows. The other options misstate row sizes and conflate storage with security or ACID.

    Read the full bite: What is a database page?

22. Question 22 of 30

    Why would an application use a presigned URL instead of proxying a private file download through its own server?

    • aTo provide permanent, unrestricted access to the file.
    • bTo enable complex, multi-file access control for different user roles.
    • cTo allow the client to authenticate directly with the cloud provider.
    • dTo reduce server load and improve efficiency by offloading file transfer.
    Show the answer

    Answer: d · To reduce server load and improve efficiency by offloading file transfer.

    The card states that proxying large files through the server is "inefficient and costly" and that presigned URLs allow the browser to download "directly from cloud storage," meaning the server doesn't handle the file stream, thus reducing server load. Option C is incorrect because the client doesn't authenticate with the cloud provider; the presigned URL itself contains the necessary authentication for that specific request.

    Read the full bite: Presigned URLs: Temporary Access to Private Files

23. Question 23 of 30

    What fundamental problem does a storage gateway primarily solve in a hybrid cloud setup?

    • aProviding a direct, unmediated connection to cloud storage for performance-critical workloads.
    • bGuaranteeing real-time, low-latency access to cloud data for all applications.
    • cEliminating the need for any local storage infrastructure.
    • dBridging the protocol gap between on-premise applications and cloud storage APIs.
    Show the answer

    Answer: d · Bridging the protocol gap between on-premise applications and cloud storage APIs.

    A storage gateway's core function is to translate traditional on-premise storage protocols (like NFS or iSCSI) into cloud-native API calls (like REST), allowing legacy applications to use cloud storage without modification. It does not eliminate all local storage, nor is it suitable for latency-sensitive applications.

    Read the full bite: Storage Gateway: Your On-Prem to Cloud Translator

24. Question 24 of 30

    Why can inserting rows with random primary-key values degrade a clustered (index-organized) table more than a heap?

    • aClustered tables must place each row in key order, causing mid-tree page splits and fragmentation
    • bHeaps store data in key order while clustered tables do not
    • cRandom keys disable the buffer pool for clustered tables
    • dHeaps reject random keys, forcing sequential inserts
    Show the answer

    Answer: a · Clustered tables must place each row in key order, causing mid-tree page splits and fragmentation

    A clustered table physically orders rows by key, so random keys force insertions into the middle and trigger page splits, while a heap simply appends. The last option reverses the definitions; heaps are unordered.

    Read the full bite: Heap file versus clustered index

25. Question 25 of 30

    What is the key security benefit of attaching a role to a VM instead of placing access keys on it?

    • aIt grants the VM unrestricted access so policies are unnecessary
    • bThe VM receives short-lived, auto-rotating credentials with no stored secret
    • cIt makes the storage bucket publicly readable for convenience
    • dIt permanently caches one static credential that never needs rotation
    Show the answer

    Answer: b · The VM receives short-lived, auto-rotating credentials with no stored secret

    An attached role delivers temporary, auto-rotating credentials via the metadata service, so no long-lived secret exists to leak. The other options either remove access control or reintroduce the static-key risk.

    Read the full bite: Granting a VM scoped storage access without static keys

26. Question 26 of 30

    After locating the leaf page holding the lower bound, how does a B+ Tree efficiently return the rest of a range?

    • aIt follows the sorted linked list between leaf pages, reading consecutive leaves until the upper bound
    • bIt loads the entire tree into memory and filters
    • cIt re-traverses from the root once per matching row
    • dIt switches to a hash lookup for each remaining key
    Show the answer

    Answer: a · It follows the sorted linked list between leaf pages, reading consecutive leaves until the upper bound

    Leaf pages are linked in sorted order, so the engine simply walks that chain from the start key onward, avoiding repeated root descents. Re-traversing per row or loading the whole tree would be far more expensive.

    Read the full bite: B+ Tree range queries across pages

27. Question 27 of 30

    A 10,000-GPU training cluster expects a hardware failure roughly every nine hours. Which checkpointing approach best balances throughput protection with recovery reliability?

    • aOn-demand checkpoints triggered only when a node failure is detected to minimize I/O
    • bHourly blocking writes to a parallel filesystem with post-training checksum verification
    • cAsynchronous checkpoints every tens of minutes using atomic finalization and tiered storage
    • dSynchronous daily checkpoints to a shared filesystem to ensure consistency across all ranks
    Show the answer

    Answer: c · Asynchronous checkpoints every tens of minutes using atomic finalization and tiered storage

    Asynchronous writes prevent GPU stalls, an MTBF-driven cadence of tens of minutes bounds lost work to an acceptable window, and atomic finalization with tiered storage guarantees valid recoverable states. Option D is a common anti-pattern: synchronous daily checkpoints destroy throughput and risk losing a full day of compute, which is economically untenable at this scale.

    Read the full bite: Robust checkpointing strategy for multi-day training jobs and seamless resumption

28. Question 28 of 30

    Why would a database administrator choose different storage engines for various tables within the same database?

    • aTo allow direct user interaction with the underlying data files for specific tables.
    • bTo ensure all data is stored in a single, unified file format for easier backups.
    • cTo optimize each table's performance based on its specific data access patterns and workload.
    • dTo simplify the process of defining table relationships and constraints.
    Show the answer

    Answer: c · To optimize each table's performance based on its specific data access patterns and workload.

    The card states that "Different engines optimize for different tasks" and provides the example of MySQL using InnoDB for transactional safety and MyISAM for faster reads, indicating engine choice is driven by workload optimization. Option D describes a logical database design function, not a storage engine's primary role.

    Read the full bite: Storage Engine: The Database's Filing System

29. Question 29 of 30

    A business intelligence team needs to frequently run queries that calculate the average order value across all customers and product types. Which database storage organization would be most efficient for this task?

    • aColumn-oriented, as it allows the database to read only the necessary columns (e.g., order value, customer ID, product type) across many rows, reducing I/O.
    • bColumn-oriented, but only if the data is frequently updated, ensuring the latest values are always available.
    • cRow-oriented, because it stores all data for a complete order together, which speeds up any query involving orders.
    • dRow-oriented, because it's optimized for transactional workloads where data consistency is paramount.
    Show the answer

    Answer: a · Column-oriented, as it allows the database to read only the necessary columns (e.g., order value, customer ID, product type) across many rows, reducing I/O.

    Column-oriented storage is designed for Online Analytical Processing (OLAP) tasks like calculating averages over many records, as it stores column data contiguously, allowing the system to read only the specific columns required, thus minimizing disk I/O. Row-oriented storage, while efficient for retrieving entire records, would be inefficient here because it would force the system to read all columns for every row, even if only a few are needed for the calculation.

    Read the full bite: Row vs. Columnar Storage: Organizing Data for Speed

30. Question 30 of 30

    What is the primary reason database engines utilize fixed-size pages for data storage?

    • aTo guarantee the physical order of rows on disk for faster sequential access.
    • bTo optimize disk I/O by performing fewer, larger read and write operations.
    • cTo simplify the process of encrypting data blocks for enhanced security.
    • dTo enable dynamic resizing of individual rows within a table without fragmentation.
    Show the answer

    Answer: b · To optimize disk I/O by performing fewer, larger read and write operations.

    The card states that pages exist to "minimize the number of slow disk operations" by grouping data into "larger, more efficient chunks" for I/O. Option A is incorrect because the card explicitly mentions "the physical order of rows on a page is not guaranteed."

    Read the full bite: Database Pages: The Building Blocks of Your Data