AWS Storage Fundamentals: S3, EFS, and FSx

Purpose of This Lesson

This lesson establishes the differences between AWS object storage and file storage, then maps common workload requirements to Amazon S3, Amazon EFS, and Amazon FSx. The main assessment skill is recognizing whether an application needs an API-accessed object store or a mounted file system, and then selecting the appropriate AWS service.

Key Concepts

Object storage versus file storage

Object storage stores data as objects inside buckets. Applications access the data through a REST API over HTTP rather than mounting the bucket as an operating-system drive. Object storage uses a flat namespace.

File storage organizes data in directories and a hierarchy. A client mounts the file system using a file-sharing protocol, after which the operating system can read and write files as though the storage were a local or network drive.

Characteristic	Object storage	File storage
AWS example	Amazon S3	Amazon EFS or Amazon FSx
Organization	Flat namespace; prefixes can imitate folders	Directories and hierarchical paths
Access	REST API over HTTP	Mounted file system using a file protocol
Operating-system mount	Not directly mountable as a normal local drive	Designed to be mounted by clients
Typical access model	Application manipulates objects	Applications use normal file operations

Amazon S3 fundamentals

Amazon S3 is an object-based storage service:

• A bucket is the container for objects.
• An object contains a key, value, version ID, metadata, and subresources. Access-control information determines what actions are permitted.
• The key is the object name, including any apparent folder path.
• The value is the actual stored data.
• Objects can be files such as documents, PDFs, videos, or images.
• An object can range from zero bytes to 5 TB.
• S3 provides unlimited overall storage capacity.
• Bucket names use a universal namespace and therefore must be globally unique.
• A bucket is created in an AWS Region, and its data is stored within that Region.
• S3 provides strong read-after-write consistency.
• Objects are accessed through REST API endpoints using HTTP methods such as GET, PUT, POST, SELECT, and DELETE.

#### Buckets, keys, and prefixes

S3 does not provide a true directory hierarchy. A key such as project-alpha/documents/report.pdf is one flat object key. The apparent folders are shared prefixes used to group and organize keys. Prefixes can be nested, but buckets cannot be created inside other buckets.

A space in an object name is represented in a URL-encoded form, such as + in the URL representation shown in the lesson. The key remains the complete object name, including its prefix.

Amazon EFS fundamentals

Amazon Elastic File System is a file-based storage service that can be created within a VPC. Clients mount EFS through mount points using the NFS protocol.

Important characteristics include:

• It is intended for Linux instances; the lesson does not describe Windows support for EFS.
• Instances in multiple Availability Zones can connect to the same file system through mount points.
• Thousands of instances can connect simultaneously.
• Access can extend from another VPC through VPC peering.
• On-premises servers can connect over a VPN or AWS Direct Connect connection.
• Once mounted, the operating system can use normal file and directory operations.

Amazon FSx fundamentals

Amazon FSx provides fully managed third-party file systems. The lesson focuses on two options:

#### FSx for Windows File Server

Use FSx for Windows File Server for Windows-based applications that need a Windows file system. Windows instances can connect from multiple Availability Zones. A Multi-AZ standby can also be created in another Availability Zone. Authentication uses AWS Managed Microsoft AD. On-premises connectivity is supported through VPN or Direct Connect.

#### FSx for Lustre

FSx for Lustre is a high-performance file system for compute-intensive workloads, including:

• Machine learning
• High-performance computing
• Video processing
• Financial modeling
• Electronic design automation

It provides a POSIX-compliant file system interface and works natively with S3. S3 objects can be presented as files through FSx for Lustre, processed by compute instances, and written back to S3 as result data. This combination can support cloud bursting and data migration scenarios. The file system can also be accessed from on-premises environments using VPN or Direct Connect.

Storage Service Selection and Workload Context

Choose S3 when the application needs object storage

S3 is appropriate when data is naturally handled as independent objects and the application can use HTTP-based APIs. It is not the right choice when an application requires a conventional mounted drive or native directory and file operations through the operating system.

Typical source-supported examples include documents, PDFs, videos, and images.

Choose EFS when Linux workloads need shared mounted storage

EFS is the relevant choice when multiple Linux instances need to mount and use the same file system through NFS. Its multi-Availability Zone connectivity and support for VPC, VPN, and Direct Connect access make it suitable for shared Linux file storage across distributed environments.

Choose FSx for Windows File Server when Windows file semantics are required

FSx for Windows File Server is the better fit for Windows applications that need a managed Windows file system and authentication through AWS Managed Microsoft AD. Consider its Multi-AZ standby capability when the scenario emphasizes resilient Windows file storage.

Choose FSx for Lustre for high-performance processing with S3 data

FSx for Lustre is designed for high-throughput, compute-intensive processing. Its decisive differentiator is the ability to expose S3 objects through a POSIX-compliant file system interface and write processed results back to S3.

Exam- or Assessment-Relevant Takeaways

• Do not confuse a bucket with a mounted file share. S3 is accessed through REST APIs; EFS and FSx are mounted file systems.
• If the scenario says Linux, NFS, and shared access from many instances, evaluate EFS.
• If the scenario says Windows application, Windows file system, or AWS Managed Microsoft AD, evaluate FSx for Windows File Server.
• If the scenario says machine learning, HPC, video processing, or another compute-intensive workload combined with S3 data, evaluate FSx for Lustre.
• S3 “folders” are prefixes in object keys, not actual directories.
• S3 bucket names must be globally unique, even though the bucket is created in a selected AWS Region.
• EFS supports Linux clients in the lesson’s scope; FSx for Windows File Server addresses Windows workloads.
• FSx for Lustre is the file-system choice that provides native S3 integration and a POSIX-compliant interface.
• For external or on-premises access, identify whether the scenario uses VPN or Direct Connect rather than assuming all storage services are publicly mounted.

Tool / Feature Decision Guide

Requirement	Recommended service	Decisive reason
Store files as independently addressable objects through HTTP APIs	Amazon S3	Object storage with REST API access
Organize data with real directories and mount it to Linux instances	Amazon EFS	Managed NFS file storage for Linux
Share file storage with Windows applications	FSx for Windows File Server	Managed Windows file system and AWS Managed Microsoft AD authentication
Process S3 data through a high-performance POSIX file system	FSx for Lustre	Presents S3 objects as files and supports results written back to S3
Provide shared access from Linux instances across Availability Zones	Amazon EFS	One file system can be mounted by instances across multiple Availability Zones
Connect on-premises systems to managed file storage	EFS or FSx, according to workload	VPN or Direct Connect provides the network path; workload operating system and file-system needs select the service

Common Traps / Misconceptions

• Treating S3 like a network drive: S3 is not mounted as a normal operating-system file system in this lesson. Applications use its REST API.
• Assuming S3 folders are real directories: A folder-looking path is only a shared prefix within an object key.
• Thinking buckets can be nested: Buckets cannot be created inside other buckets.
• Confusing the S3 key and value: The key is the object name; the value is the stored data.
• Assuming a bucket name only needs to be unique within its Region: S3 bucket names use a global namespace.
• Selecting EFS for Windows instances: The lesson identifies EFS as Linux-only and directs Windows file workloads to FSx for Windows File Server.
• Choosing FSx for Lustre merely because it is a file system: Its strongest fit is high-performance, compute-intensive processing, especially when S3 integration is valuable.
• Assuming all file storage is automatically reachable from on premises: The environment still needs connectivity through VPN or Direct Connect.

Real-World Engineer / Analyst Notes

• Start storage analysis with the application’s access pattern, not with the file extension. A PDF can be stored in S3, EFS, or FSx; the correct choice depends on whether the application needs object API access or mounted file semantics.
• Verify the required client operating system and protocol early. Linux/NFS points toward EFS, while Windows file requirements point toward FSx for Windows File Server.
• Keep S3 data near its users when latency matters, while remembering that the bucket is associated with an AWS Region.
• Use S3 prefixes for logical organization, but do not design operations that depend on directory behavior that object storage does not provide.
• For processing pipelines, FSx for Lustre can bridge the access models: compute uses a POSIX file interface while source and result data remain in S3.
• In hybrid designs, validate the VPN or Direct Connect path and ensure that the selected file system is reachable from the intended VPC or on-premises environment.

Quick Reference Summary

• S3: Object storage; buckets contain objects; REST API over HTTP; flat namespace; prefixes imitate folders; globally unique bucket names; object size from 0 bytes to 5 TB; unlimited overall storage; strong read-after-write consistency.
• EFS: Managed file storage for Linux; NFS mounts; shared access by thousands of instances; supports multi-Availability Zone use, VPC peering, VPN, and Direct Connect.
• FSx for Windows File Server: Managed Windows file storage; supports Windows workloads, Multi-AZ standby, AWS Managed Microsoft AD, VPN, and Direct Connect.
• FSx for Lustre: High-performance POSIX file system for compute-intensive workloads; integrates natively with S3 and can return processing results to S3.
• Primary decision: Choose object storage when the application uses APIs and objects; choose file storage when the operating system must mount and manipulate a shared hierarchy.

Flashcards

Q: An application needs to store PDFs and video files and access them through HTTP API calls rather than a mounted drive. Which service best fits?
A: Amazon S3, because it is object storage accessed through REST APIs over HTTP.

Q: What is the difference between an S3 object key and its value?
A: The key is the object’s name, including any apparent prefix path. The value is the actual stored data.

Q: A team says it created an S3 folder named project-alpha. What does that folder actually represent?
A: It is a shared prefix in object keys, not a real directory in a hierarchical file system.

Q: Why must S3 bucket names be globally unique even when buckets are created in Regions?
A: The bucket name forms part of the S3 URL and uses a universal namespace, so uniqueness is global rather than limited to one Region.

Q: A large group of Linux instances in multiple Availability Zones must mount and share the same file system. Which service should be evaluated first?
A: Amazon EFS, because it provides NFS-based file storage that can be mounted by Linux instances across Availability Zones.

Q: When would you select EFS instead of S3?
A: Select EFS when the workload needs a mounted file system, directories, and normal operating-system file operations. Select S3 when the application can work with objects through REST APIs.

Q: What protocol does the lesson associate with EFS access?
A: NFS. EFS file systems are mounted by clients using NFS.

Q: A Windows application needs managed shared file storage and authentication through AWS Managed Microsoft AD. Which FSx option fits?
A: Amazon FSx for Windows File Server, because it targets Windows file workloads and uses AWS Managed Microsoft AD for authentication.

Q: Which FSx option is intended for machine learning, HPC, video processing, and financial modeling?
A: FSx for Lustre, which is optimized for high-performance, compute-intensive workloads.

Q: What makes FSx for Lustre useful in an S3-based processing pipeline?
A: It presents S3 objects as files through a POSIX-compliant file-system interface and can write processing results back to S3.

Q: An on-premises server must access EFS or FSx. What connectivity options identified in the lesson can provide the connection?
A: A VPN or AWS Direct Connect connection can connect the on-premises environment to the AWS network hosting the file system.

Q: What is the S3 object-size range stated in the lesson?
A: An individual S3 object can be from zero bytes to 5 TB.

Practice Questions

Question 1

A company has a Linux application running on instances in several Availability Zones. All instances need simultaneous read and write access to the same directory hierarchy using standard file operations. Which service is the best fit?

A. Amazon S3
B. Amazon EFS
C. Amazon FSx for Windows File Server
D. Amazon FSx for Lustre

Correct answer: B. Amazon EFS

Explanation: The decisive clues are Linux clients, a mounted directory hierarchy, NFS-style file access, and shared use across Availability Zones. S3 is API-based object storage, while FSx for Windows is intended for Windows workloads.

Question 2

A machine learning pipeline must process objects stored in S3 using a high-performance POSIX-compliant file system and then write the results back to S3. Which service should the engineer choose?

A. Amazon EFS
B. Amazon S3 only
C. Amazon FSx for Lustre
D. Amazon FSx for Windows File Server

Correct answer: C. Amazon FSx for Lustre

Explanation: FSx for Lustre is designed for compute-intensive processing and natively presents S3 objects as files while supporting results written back to S3.

Question 3

An engineer creates an object with the key finance/2026/report.pdf and concludes that S3 has created a finance directory containing the report. What is the accurate interpretation?

A. S3 created a nested bucket called finance.
B. S3 created a true directory hierarchy inside the bucket.
C. finance/ is a shared prefix that forms part of the object’s flat key.
D. The object is stored outside the bucket because folders are separate resources.

Correct answer: C. finance/ is a shared prefix that forms part of the object’s flat key.

Explanation: S3 uses a flat namespace. Folder-like organization is produced by prefixes in object keys; buckets cannot contain other buckets.

Question 4

A Windows-based application requires a managed file system, Windows clients in multiple Availability Zones, and authentication through AWS Managed Microsoft AD. Which option is most appropriate?

A. Amazon S3
B. Amazon EFS
C. Amazon FSx for Windows File Server
D. Amazon FSx for Lustre

Correct answer: C. Amazon FSx for Windows File Server

Explanation: Windows file-system requirements and AWS Managed Microsoft AD identify FSx for Windows File Server. EFS is described as Linux-only, and FSx for Lustre targets high-performance processing workloads.

WordPress Metadata

Suggested Slug:
aws-storage-s3-efs-fsx-fundamentals

Meta Description:
Compare Amazon S3 object storage with Amazon EFS and Amazon FSx file storage, including access methods, workload fit, connectivity, and key CloudOps decisions.

Tags:
AWS, Amazon S3, Amazon EFS, Amazon FSx, Object Storage, File Storage, NFS, SMB, Amazon FSx for Lustre, AWS Certified CloudOps Engineer