Object vs block vs file storage

WHAT THIS TESTS The interviewer is checking that you understand the three storage models and instinctively pick the right one for a workload.

A GOOD ANSWER COVERS Object storage, such as S3, Blob, or Cloud Storage, keeps data as objects, each a blob plus metadata and a unique key, in a flat namespace accessed over HTTP. It scales nearly infinitely and is cheap and durable, but objects are written and read whole rather than edited in place, making it ideal for images, video, backups, logs, and static web assets. Block storage, such as EBS or Persistent Disk, exposes a raw volume the operating system formats and mounts, offering low-latency random read and write to a single attached instance, which suits databases and boot disks. File storage, such as EFS or Filestore, provides a shared hierarchical file system multiple instances can mount over NFS or SMB at once, suited to shared content, home directories, or lift-and-shift apps expecting a POSIX file system.

COMMON WRONG ANSWERS Using object storage as a database, expecting in-place updates and low-latency random access. Trying to share one block volume across many instances simultaneously. Confusing file storage's shared semantics with block storage's single-attachment model. Assuming object storage offers a real directory hierarchy rather than a flat keyspace with prefix conventions.

LIKELY FOLLOW-UPS Why can object storage not back a relational database. How do you share storage across instances. What are the durability and consistency differences. How do prefixes simulate folders in object storage.

ONE CONCRETE EXAMPLE A photo-sharing app stores uploaded images and thumbnails in object storage for cheap, durable, web-served access, runs its Postgres database on a block volume for fast transactional reads and writes, and mounts a shared file system across its rendering fleet so every worker reads the same set of font and template files. Each storage type is matched to the access pattern it was built for.