AWS Storage Services Strategy: S3, EFS, EBS, and FSx

This guide provides a comprehensive breakdown of AWS storage solutions tailored for the Solutions Architect Professional (SAP-C02) exam, focusing on architectural selection, performance characteristics, and migration strategies.

Learning Objectives

After studying this guide, you should be able to:

• Differentiate between block (EBS), file (EFS/FSx), and object (S3) storage models.
• Select the appropriate FSx flavor based on protocol (SMB vs. NFS) and workload requirements (HPC vs. Enterprise Migration).
• Architect Multi-AZ shared storage solutions while identifying OS-specific limitations.
• Recommend data migration tools like AWS DataSync and AWS Transfer Family for hybrid cloud scenarios.

Key Terms & Glossary

• IOPS (Input/Output Operations Per Second): A performance metric used to measure the speed of storage devices (critical for EBS and FSx for OpenZFS).
• NFS (Network File System): A protocol used by Linux/macOS for shared file access; primary protocol for Amazon EFS.
• SMB (Server Message Block): A protocol primarily used by Windows for file sharing; supported by FSx for Windows and NetApp ONTAP.
• POSIX Compliance: A set of standard operating system interfaces; required for many legacy Linux applications moving to EFS or FSx.
• Multi-Attach: An EBS feature allowing a single volume to be attached to multiple EC2 instances (same AZ only).

The "Big Idea"

At the Professional level, storage is not just about where data sits; it is about Access Patterns and Connectivity. The architecture choice is driven by two questions: Who needs the data (one instance or many?) and How do they speak to it (Block, File, or API?). Success on the exam depends on knowing when to move from simple block storage to shared file systems to enable scalability and high availability.

Formula / Concept Box

Hierarchical Outline

• I. Block Storage: Amazon EBS
  • Scope: Bound to a single Availability Zone (AZ).
  • Multi-attach: Max 16 Linux instances; requires Provisioned IOPS (io1/io2).
  • Best Use: Boot volumes, transactional databases, 1-to-1 instance mapping.
• II. Shared File Storage: Amazon EFS
  • Protocols: NFSv4 only (No Windows support).
  • Storage Classes: Standard and One Zone (for cost-saving without Multi-AZ durability).
  • Performance: Elastic throughput that scales with data size.
• III. Specialized File Storage: Amazon FSx Family
  • FSx for NetApp ONTAP: Best for on-prem migrations (supports NFS, SMB, iSCSI).
  • FSx for Windows: Native SMB support with Active Directory integration.
  • FSx for Lustre: High-performance computing (HPC) and machine learning.
  • FSx for OpenZFS: High-throughput Linux workloads.
• IV. Migration Services
  • AWS DataSync: Agent-based for on-prem; agentless for AWS-to-AWS transfers.
  • AWS Transfer Family: Managed FTP, SFTP, and FTPS gateway for S3/EFS.

Visual Anchors

Storage Decision Flow

Multi-AZ File Architecture

\begin{tikzpicture}[font=\small] % Draw AZs \draw[dashed, gray] (0,0) rectangle (4,4) node[above] {AZ-A}; \draw[dashed, gray] (5,0) rectangle (9,4) node[above] {AZ-B};

% Draw Instances \draw[fill=orange!20] (1,1) rectangle (3,2) node[midway] {EC2 Instance}; \draw[fill=orange!20] (6,1) rectangle (8,2) node[midway] {EC2 Instance};

% Draw EFS \draw[fill=blue!10, thick] (2, -1) rectangle (7, -0.5) node[midway] {Amazon EFS (Regional)};

% Draw connections \draw[<->, thick, blue] (2,1) -- (2,-0.5); \draw[<->, thick, blue] (7,1) -- (7,-0.5);

% Labels \node[text width=4cm, align=center] at (4.5, -1.5) {Shared File System \ (Multi-AZ Mount Targets)}; \end{tikzpicture}

Definition-Example Pairs

• EBS (Elastic Block Store)
  • Definition: High-performance block storage for use with EC2.
  • Example: Running a legacy SQL Server database that requires sub-millisecond latency and specific disk formatting.
• EFS (Elastic File System)
  • Definition: Serverless, fully managed NFS file system.
  • Example: A fleet of web servers sharing a common directory for uploading and serving images.
• AWS Transfer Family
  • Definition: A gateway for transferring files via FTP protocols into S3 or EFS.
  • Example: A third-party vendor needs to upload CSV files via SFTP, but you want to process those files using AWS Lambda in S3.

Worked Examples

Example 1: The Migration Choice

Scenario: A company is migrating a large Linux-based data analytics application from an on-premises NetApp storage array to AWS. They require NFS and iSCSI support. Solution: Select Amazon FSx for NetApp ONTAP. Reasoning: While EFS supports NFS, it does not support iSCSI. FSx for NetApp ONTAP is specifically designed for these migrations, providing the same features as on-premises NetApp arrays.

Example 2: High-Performance Computing

Scenario: A financial firm needs to run massive simulations across hundreds of Linux instances. The storage must handle hundreds of GB/s of throughput. Solution: Amazon FSx for Lustre. Reasoning: Lustre is optimized for compute-intensive workloads where throughput and speed are the primary requirements, outperforming EFS in high-parallelism scenarios.

Comparison Tables

FSx Flavor Comparison

Checkpoint Questions

1. Which storage service should be used if you need a shared file system for Windows instances that spans multiple AZs?
2. True or False: EBS Multi-Attach allows an io2 volume to be shared between instances in different AZs.
3. What service provides an agentless way to move data between S3 and EFS?
4. If a client needs to access S3 data using SFTP, which service is the best fit?

▶Click for Answers

1. Amazon FSx for Windows File Server (or FSx for NetApp ONTAP).
2. False. Multi-attach is limited to a single AZ.
3. AWS DataSync (AWS-to-AWS is agentless).
4. AWS Transfer Family.

Muddy Points & Cross-Refs

• EBS Multi-attach vs. EFS: Students often confuse these. Remember: EBS Multi-attach is for Block access in one AZ. EFS is for File access across multiple AZs.
• One-Zone vs. Standard: EFS One-Zone is cheaper but loses the 3-AZ durability. Use it for development or easily reproducible data.
• FSx for OpenZFS: Newest member. Think of it as the high-speed version of EFS for specific Linux workloads that need ZFS features like snapshots or cloning.