Core Storage Services: S3, EBS, EFS, and Glacier

The Cloud Practitioner exam expects you to match a storage service to a use case in seconds. The key is recognizing three storage types — object, block, and file — and then knowing which AWS service delivers each, plus how S3's storage classes trade cost against retrieval speed. Learn the decision cues below and most storage questions become quick eliminations.

Core Idea

• Three storage types: object storage (whole files with metadata, accessed over HTTP) is S3; block storage (raw volumes attached to one server) is EBS; shared file storage (mounted by many servers) is EFS.
• S3 is regional, EBS is tied to one Availability Zone, and EFS is regional and shared across many EC2 instances at once.
• Match cost to access pattern: frequently accessed data goes to hot classes; rarely touched data goes to Infrequent Access or Glacier archival tiers to cut cost.

Object vs. Block vs. File Storage

• Object storage (Amazon S3): stores each file as a whole object with a unique key and metadata in a flat namespace. You cannot edit part of an object — you replace the whole thing. Accessed via API/HTTP, not mounted as a drive. Ideal for images, backups, static websites, data lakes, and logs.
• Block storage (Amazon EBS): presents raw blocks like a physical hard drive. The OS puts a file system on top, so you can change individual blocks. Attaches to a single EC2 instance. Ideal for boot volumes, databases, and any app needing low-latency read/write.
• File storage (Amazon EFS / FSx): a shared file system mounted by many clients at once using standard file paths and permissions. Ideal for shared content, web server fleets, and lift-and-shift workloads needing a common mount point.

Amazon S3

S3 is object storage where files live in buckets (globally unique names) as objects identified by keys. Highlights:

• Durability of 99.999999999% (eleven 9s) — data is automatically replicated across multiple Availability Zones in a Region (except One Zone classes).
• Virtually unlimited capacity; you pay for what you store and for requests/transfer.
• Versioning keeps multiple versions of an object so you can recover from accidental overwrites or deletes.
• Lifecycle policies automatically transition objects to cheaper classes or expire them over time.
• Common uses: backups, static website hosting, media, big-data lakes, and application assets.

S3 Storage Classes

Classes trade lower storage price for higher retrieval cost, slower retrieval, or less redundancy:

• S3 Standard — frequently accessed data; highest storage cost, no retrieval fee, multi-AZ.
• S3 Standard-Infrequent Access (Standard-IA) — long-lived but rarely accessed data; lower storage cost, but a per-GB retrieval fee. Multi-AZ.
• S3 One Zone-IA — like Standard-IA but stored in a single AZ; cheaper, but data is lost if that AZ is destroyed. Use for reproducible/secondary copies.
• S3 Intelligent-Tiering — automatically moves objects between tiers based on access patterns for a small monitoring fee; best when access is unknown or changing.
• Glacier Instant Retrieval — archive with millisecond access; cheap storage for rarely accessed data you still need instantly.
• Glacier Flexible Retrieval — archive with retrieval in minutes to hours; very low storage cost.
• Glacier Deep Archive — lowest cost of all; retrieval takes hours (up to ~12). Ideal for long-term compliance/regulatory archives kept for years.

EBS, Instance Store, EFS, and Storage Gateway

• Amazon EBS: durable block volumes attached to one EC2 instance in the same AZ; persists independently of the instance. Snapshots back up a volume to S3 and can restore or copy volumes across AZs/Regions.
• Instance store: temporary block storage physically attached to the host. Very fast but ephemeral — data is lost when the instance stops or terminates. Use for caches or scratch data only.
• Amazon EFS: elastic, fully managed shared file system for Linux; automatically grows/shrinks and can be mounted by thousands of instances across multiple AZs at once.
• AWS Storage Gateway: hybrid service connecting on-premises apps to AWS cloud storage (file, volume, tape gateways), useful for backup, migration, and bursting to the cloud.

High-Yield Exam Patterns

• "Object storage" or "store files/images/backups" → S3. "Attached to an EC2 instance like a disk" → EBS. "Shared file system for many instances" → EFS.
• "Rarely accessed but must be retrieved instantly" → Glacier Instant Retrieval; "long-term archive, hours to retrieve, lowest cost" → Glacier Deep Archive.
• "Access pattern is unknown or unpredictable" → S3 Intelligent-Tiering.
• "Reproducible data, save cost, okay if one AZ fails" → S3 One Zone-IA.
• Eleven 9s of durability is an S3 signature phrase; snapshots are the EBS backup mechanism (stored in S3).
• "Temporary, high-speed, data can be lost" → instance store; "connect on-premises to cloud storage" → Storage Gateway.

Common Traps to Avoid

• Confusing durability (eleven 9s — data won't be lost) with availability (uptime access) — they are different metrics.
• Thinking EBS can attach to many instances at once — a standard EBS volume attaches to one instance in its AZ; use EFS for shared multi-instance access.
• Assuming instance store persists — it is ephemeral and lost on stop/terminate.
• Choosing Standard-IA for frequently accessed data — the per-retrieval fee makes it more expensive when access is high.
• Forgetting that S3 bucket names are globally unique and objects are replaced whole, not edited in place.