AWS Storage Services: Architecture and Use Cases

This guide covers the core AWS storage portfolio including object, block, and file storage, providing the architectural context needed for the AWS Solutions Architect Associate (SAA-C03) exam.

Learning Objectives

• Differentiate between block, file, and object storage architectures.
• Identify appropriate use cases for Amazon S3, EBS, EFS, and FSx.
• Evaluate performance requirements (IOPS vs. Throughput) to select the correct EBS volume type.
• Compare shared storage options for Windows and Linux environments.
• Select cost-optimization strategies using S3 Lifecycle policies and storage tiering.

Key Terms & Glossary

• Object Storage: Data managed as discrete units (objects) with metadata and a unique identifier. Ideal for unstructured data.
• Block Storage: Data split into fixed-size blocks; accessed by an operating system as a local disk drive (e.g., EBS).
• File Storage (NAS): Data stored in a hierarchy of folders and files; accessible via network protocols like NFS or SMB.
• IOPS (Input/Output Operations Per Second): A measure of the number of reads/writes per second—critical for database performance.
• Durability: The probability that data will not be lost (S3 offers 99.999999999% durability).
• Throughput: The amount of data transferred per second (e.g., MB/s)—critical for big data and streaming.

The "Big Idea"

In the AWS ecosystem, storage is not "one size fits all." The architectural decision depends entirely on the access pattern. If a single server needs a high-performance boot disk, use Block (EBS). If thousands of clients need to access the same unstructured images, use Object (S3). If multiple Linux servers need a shared configuration directory, use File (EFS). Choosing the wrong type leads to either performance bottlenecks or unnecessary costs.

Formula / Concept Box

[!NOTE] Note: EBS Multi-Attach is possible for specific IOPS-optimized volumes, but generally, EBS is one-to-one.

Hierarchical Outline

1. Object Storage (Amazon S3)
   • Buckets and Objects: Global namespace for buckets; regional storage.
   • Storage Classes: Standard, IA (Infrequent Access), One Zone-IA, and Glacier (Archive).
   • Features: Versioning, Replication (CRR/SRR), and Lifecycle Policies.
2. Block Storage (Amazon EBS)
   • SSD-backed (gp3, io2): Best for transactional workloads and databases.
   • HDD-backed (st1, sc1): Best for large sequential workloads (throughput-focused).
   • Snapshots: Point-in-time incremental backups stored in S3.
3. File Storage (EFS & FSx)
   • Amazon EFS: Fully managed NFS for Linux; regional availability (Multi-AZ).
   • Amazon FSx for Windows: Native SMB support, integrates with Active Directory.
   • Amazon FSx for Lustre: High-performance for compute-heavy (ML, HPC) tasks.
4. Hybrid & Migration
   • Storage Gateway: Connects on-premises to cloud storage.
   • DataSync: High-speed data transfer service.

Visual Anchors

Storage Decision Tree

Instance and EBS Relationship

Definition-Example Pairs

• Amazon S3 (Standard): Object storage for frequently accessed data.
  • Example: Storing user profile pictures for a social media app that are served via a CDN.
• Amazon EBS (Provisioned IOPS): High-performance block storage.
  • Example: Hosting a mission-critical MongoDB or SQL Server database that requires 16,000 stable IOPS.
• Amazon FSx for Lustre: A distributed file system for sub-millisecond latencies.
  • Example: Feeding massive datasets into a cluster of GPU instances for machine learning model training.
• S3 Glacier Deep Archive: Lowest-cost storage class for long-term retention.
  • Example: Storing 10 years of financial audit logs that are only retrieved once a year with a 12-hour lead time.

Worked Examples

Example 1: Shared Configuration for Autoscale Group

Scenario: A company has a fleet of Linux web servers in an Auto Scaling Group. They need to share a common set of images and code configuration files that change frequently.

• Solution: Use Amazon EFS.
• Why: EFS supports the NFS protocol, allowing multiple EC2 instances to mount the same file system simultaneously across different Availability Zones. Unlike S3, it behaves like a standard directory on the OS.

Example 2: Migrating a Windows File Share

Scenario: An enterprise needs to move their 50TB Windows file server to AWS. It must support NTFS permissions and integrate with their existing Active Directory.

• Solution: Amazon FSx for Windows File Server.
• Why: It is built on Windows Server and natively supports SMB, NTFS, and AD integration, making the migration seamless for end-users who expect a mapped Z:\ drive.

Checkpoint Questions

1. Which storage service should you use for a high-performance database requiring sub-millisecond latency and local block access?
2. What is the main difference between S3 Standard and S3 Standard-IA?
3. You have a Linux-based HPC workload that requires hundreds of GB/s throughput. Which FSx flavor is best?
4. True or False: EBS volumes are automatically replicated across multiple Availability Zones.
5. How can you automatically move S3 objects to a cheaper storage class after 90 days?

▶Click to see answers

1. Amazon EBS (specifically Provisioned IOPS SSD).
2. S3 Standard-IA has a lower storage price but charges a retrieval fee per GB, making it better for infrequently accessed data.
3. FSx for Lustre.
4. False. EBS volumes are replicated within a single AZ. EFS and S3 are replicated across multiple AZs.
5. By using S3 Lifecycle Policies.