Storage Services (S3, EBS, EFS, Glacier)
AWS offers three main storage paradigms: object storage (S3 — unlimited scale, 11-nines durability), block storage (EBS — persistent volumes attached to EC2 within a single AZ), and file storage (EFS — shared NFS mountable by thousands of Linux instances across multiple AZs). Understanding S3 storage classes, EBS snapshot behavior, and when to use each storage type is essential for CLF-C02.
1. AWS Storage Services Overview
2. Amazon S3 — Object Storage
Core Concepts
- Buckets — Containers for objects; name must be globally unique across all AWS accounts; created in a specific Region
- Objects — Files plus metadata stored in S3; maximum object size is 5 TB
- Durability: 99.999999999% (11 nines) — designed so data is not lost
- Availability: 99.99% for Standard class — the data is accessible when you need it
- By default, all S3 buckets and objects are private — access is granted via bucket policies or IAM
S3 Storage Classes
| Storage Class | Retrieval | Min Duration | AZs | Best For |
|---|---|---|---|---|
| Standard | Milliseconds | None | 3 | Frequently accessed data, websites, analytics |
| Intelligent-Tiering | Milliseconds | None | 3 | Unknown or changing access patterns |
| Standard-IA | Milliseconds | 30 days | 3 | Backups, disaster recovery, infrequently accessed |
| One Zone-IA | Milliseconds | 30 days | 1 | Non-critical reproducible infrequent data |
| Glacier Instant | Milliseconds | 90 days | 3 | Medical images, news archives needing fast access |
| Glacier Flexible | 1 min–12 hrs | 90 days | 3 | Long-term backups, compliance archives |
| Glacier Deep Archive | 12 hours | 180 days | 3 | Regulatory data accessed once per year or less |
S3 Key Features
| Feature | Description |
|---|---|
| Versioning | Keep multiple versions of each object; protect against accidental deletion or overwrite |
| Lifecycle Policies | Automatically transition objects between storage classes or delete after a period |
| Cross-Region Replication | Automatically copy objects to another Region; requires versioning enabled |
| Pre-signed URLs | Temporary URL granting access to a private object without AWS credentials |
| Server-Side Encryption | Encrypt data at rest using SSE-S3 (AWS-managed), SSE-KMS, or SSE-C (customer key) |
| S3 Block Public Access | Account-level or bucket-level setting to prevent any public access |
| Object Lock | WORM protection — write once, read many; prevents deletion for a set period |
| Static Website Hosting | Host static HTML, CSS, JavaScript sites directly from an S3 bucket |
| Transfer Acceleration | Speed up uploads via CloudFront edge locations |
3. Amazon EBS — Block Storage
Elastic Block Store provides persistent, network-attached block storage for EC2 instances — like a virtual hard drive.
Key EBS properties:
- Same AZ required — EBS volume must be in the same AZ as the EC2 instance it attaches to
- Persists independently — data survives instance stop, start, and reboot (not termination unless configured)
- One attachment — typically one EC2 instance at a time (except io1/io2 Multi-Attach)
- Snapshots — point-in-time backups stored in S3; incremental; can copy across Regions
EBS Volume Types:
| Type | Performance | Use Case |
|---|---|---|
| gp3 / gp2 (General Purpose SSD) | Balanced IOPS and throughput | Boot volumes, dev/test, most workloads |
| io1 / io2 (Provisioned IOPS SSD) | Highest consistent IOPS | High-performance databases |
| st1 (Throughput Optimized HDD) | High sequential throughput | Big data, log processing, data warehouses |
| sc1 (Cold HDD) | Lowest cost | Infrequently accessed cold data |
4. Instance Store — Ephemeral Storage
- Block storage that is physically attached to the EC2 host server (not network-attached)
- Extremely fast — higher throughput and lower latency than EBS because there is no network hop
- Ephemeral — data is permanently lost when the instance stops, terminates, or fails
- Cannot be detached or moved to another instance
- Use case: Temporary buffers, caches, scratch data — never for data that must persist
5. Amazon EFS — Elastic File System
EFS provides a fully managed, elastic NFS (Network File System) for Linux workloads:
| Feature | EFS | EBS |
|---|---|---|
| Storage type | File (NFS) | Block |
| Multi-instance | Thousands simultaneously | One at a time |
| AZ scope | Regional (spans multiple AZs) | Single AZ |
| Scaling | Elastic — grows and shrinks automatically | Fixed provisioned size |
| OS support | Linux only | Linux and Windows |
| Use cases | Shared content repos, CMS, home directories | Boot volumes, databases |
6. Amazon FSx — Managed Third-Party File Systems
| Service | Protocol | Use Case |
|---|---|---|
| FSx for Windows File Server | SMB | Windows workloads; Active Directory integration |
| FSx for Lustre | Lustre | High-performance computing, ML training, media processing |
| FSx for NetApp ONTAP | NFS, SMB, iSCSI | Enterprise file storage with advanced features |
| FSx for OpenZFS | NFS | ZFS-based file system with snapshots |
7. AWS Snow Family — Offline Data Transfer
When migrating large datasets where network transfer is too slow or expensive:
| Device | Capacity | Form Factor | Use Case |
|---|---|---|---|
| Snowcone | 8 TB HDD or 14 TB SSD | Backpack-portable | Remote and edge locations; small migrations |
| Snowball Edge Storage | 80 TB usable | Rugged briefcase appliance | Large datacenter migrations |
| Snowball Edge Compute | 40 TB + GPU/EC2 | Rugged appliance | Edge ML processing before shipping data |
| Snowmobile | 100 PB | 45-foot shipping container truck | Entire datacenter migration |
Rule: If transferring data over your network would take weeks → consider Snow Family.
8. AWS Storage Gateway — Hybrid Storage
Connects on-premises applications to AWS cloud storage:
| Gateway Type | Protocol | What It Does |
|---|---|---|
| File Gateway | NFS / SMB | On-premises files stored as S3 objects; local cache for low-latency access |
| Volume Gateway | iSCSI | On-premises block volumes backed by S3; cached or stored modes |
| Tape Gateway | iSCSI VTL | Replaces physical tape library with S3 and Glacier archives |
9. CLF-C02 Quick Reference
| Topic | Key Fact |
|---|---|
| S3 durability | 99.999999999% (11 nines) across 3 AZs |
| S3 max object size | 5 TB |
| S3 bucket naming | Globally unique across all AWS accounts |
| S3 default access | Private — must explicitly grant access |
| S3 Standard | Frequent access; highest cost; millisecond retrieval |
| S3 Intelligent-Tiering | Auto-moves tiers; monitoring fee; no retrieval fee |
| S3 Glacier Deep Archive | Lowest cost; 12-hr retrieval; 180-day minimum |
| S3 Versioning | Keeps all versions; protects against accidental deletion |
| EBS | Block storage; same AZ as EC2; persists on stop; one instance |
| EBS snapshots | Point-in-time; incremental; stored in S3; can copy cross-Region |
| Instance Store | Ephemeral; lost on stop/terminate; very fast; no network hop |
| EFS | Managed NFS; multi-instance mount; regional; Linux only; elastic |
| EFS vs EBS | EFS: multi-instance, regional; EBS: single instance, single AZ |
| FSx for Windows | SMB; Windows and Active Directory integration |
| Snowball Edge | 80 TB physical device for large migrations |
| Snowmobile | 100 PB container truck for datacenter migrations |
| Storage Gateway | Hybrid bridge from on-premises to S3 and Glacier |
Practice Questions10
Q1. What type of storage does Amazon S3 provide?
Select one answer before revealing.
Q2. Which S3 storage class provides the LOWEST cost for data that is accessed at most once or twice per year and can tolerate retrieval times of up to 12 hours?
Select one answer before revealing.
Q3. What is the durability guarantee of Amazon S3?
Select one answer before revealing.
Q4. A company needs block storage that persists independently after an EC2 instance is stopped or terminated and can be re-attached to a different instance. Which AWS service should they use?
Select one answer before revealing.
Q5. What is the key difference between Amazon EBS and EC2 Instance Store?
Select one answer before revealing.
Q6. A company has 200 EC2 instances spread across multiple AZs that all need to read and write to the same shared file system simultaneously. Which storage service supports this?
Select one answer before revealing.
Q7. A company needs to transfer 80 TB of on-premises data to AWS. Their internet connection is 100 Mbps and they calculate the transfer would take approximately 73 days. What is the MOST practical solution?
Select one answer before revealing.
Q8. Which AWS service connects on-premises storage environments to AWS cloud storage, supporting use cases like backup to the cloud, file sharing, and tape replacement?
Select one answer before revealing.
Q9. Which TWO Amazon S3 storage classes are specifically designed for infrequently accessed data but provide millisecond retrieval (unlike Glacier)? (Select TWO — more than one answer may be correct)
Select one answer before revealing.
Q10. Which S3 feature keeps multiple versions of the same object to protect against accidental deletion and overwrites?
Select one answer before revealing.