AWS File Storage: EFS & FSx
In the AWS ecosystem, file storage provides a hierarchical structure for data that can be accessed by multiple instances simultaneously. Unlike Block Storage (EBS), which is restricted to a single instance in most cases, File Storage is designed for shared access, scalability, and high durability across multiple Availability Zones.
The Office Analogy
Think of Amazon EFS like a Shared Office Kitchen. It’s open to everyone (multiple EC2 instances), it grows automatically as more people bring food (elastic scaling), and you only pay for the snacks currently on the shelf. Amazon FSx is like a Specialized Departmental Vault. If the Accounting team needs a specific Windows-based filing system (FSx for Windows) or the Research team needs a super-high-speed data shredder for massive computations (FSx for Lustre), you use the specialized vault designed for those specific tools.
Core Concepts & Well-Architected Framework
1. Performance Efficiency
EFS offers two performance modes: General Purpose (ideal for web servers/CMS) and Max I/O (for massive parallel workloads). FSx for Lustre provides sub-millisecond latencies for High-Performance Computing (HPC).
2. Cost Optimization
Use EFS Lifecycle Management to automatically move files not accessed for 30 days to the Infrequent Access (IA) storage class, saving up to 92% in costs.
3. Reliability
EFS is a regional service, meaning data is stored redundantly across multiple Availability Zones (AZs) by default. FSx for Windows offers Multi-AZ deployments to ensure business continuity for enterprise applications.
Service Comparison Table
| Feature | Amazon EFS | FSx for Windows | FSx for Lustre |
|---|---|---|---|
| Protocol | NFS v4.0/4.1 (Linux) | SMB (Windows) | Lustre (High Perf) |
| Scaling | Fully Elastic | Manual/Scheduled | Provisioned |
| Durability | Regional (Multi-AZ) | Single or Multi-AZ | Replicated/Scratch |
| Best For | CMS, Home Directories | Active Directory, .NET | Machine Learning, HPC |
Scenario-Based Decision Matrix
If/Then Guide
- IF you need a shared drive for thousands of Linux EC2 instances THEN use Amazon EFS.
- IF you have a legacy Windows App requiring Microsoft Active Directory THEN use FSx for Windows File Server.
- IF you need to process S3 data at lightning speeds for AI training THEN use FSx for Lustre.
- IF you need to migrate an on-premises NetApp filer to AWS THEN use FSx for ONTAP.
Exam Tips: Golden Nuggets
- EFS = Linux: If the question mentions Windows instances, EFS is almost always the wrong answer.
- S3 Integration: FSx for Lustre can “lazy load” data directly from S3 and write results back to S3.
- Security: Always use Security Groups to control NFS (Port 2049) traffic to EFS.
- Throughput: EFS “Provisioned Throughput” is for workloads with high throughput-to-storage ratios.
File Storage Architecture Visualized
Key Services
EFS: Serverless, elastic, regional storage for Linux.
FSx Windows: Native SMB support, NTFS, and AD integration.
FSx Lustre: High-performance linked with S3.
Common Pitfalls
- Using EFS for Windows (Use FSx).
- Ignoring EFS Burst Credits (Monitor via CloudWatch).
- Not configuring Security Groups for Mount Targets.
Quick Patterns
- WordPress: Store /wp-content on EFS for auto-scaling.
- VDI: FSx for Windows for User Profiles.
- Big Data: FSx Lustre for temp high-speed processing.