Amazon EFS: Regional, One Zone, Storage Classes, and Throughput – SAP-C02 Study Guide

Purpose of This Lesson

Amazon Elastic File System (Amazon EFS) is a managed, shared file system designed for Linux workloads. Multiple compute instances can mount the same file system concurrently, including instances running in different Availability Zones.

This lesson focuses on EFS deployment models, connectivity, storage classes, replication, backups, and throughput choices—common decision points in AWS architecture exams.

Key Concepts

Shared NFS file system

EFS exposes a file system through the Network File System (NFS) protocol, specifically NFS version 4. EFS is intended for Linux-based clients; it is not a native Windows file system service.

Multiple Amazon EC2 instances can mount the same EFS file system and access shared files. EFS supports file locking for read and write operations, helping coordinate concurrent access.

Regional file systems

A regional EFS file system is designed for multi-AZ availability within an AWS Region. It uses mount targets associated with Availability Zones. These mount targets are represented by elastic network interfaces in the VPC.

For best performance and availability, an instance should mount the EFS mount target in its local Availability Zone. This avoids unnecessary cross-AZ traffic and reduces latency.

Operations on a regional file system are durably stored across Availability Zones, providing resilience against the loss of an individual Availability Zone.

One Zone file systems

An EFS One Zone file system stores data in a single Availability Zone and has a mount target in that Availability Zone. It can be less expensive than a regional file system, but it does not provide the same multi-AZ resilience.

A client in another Availability Zone can connect to the One Zone file system through its mount target, but this introduces cross-AZ access considerations and does not change the underlying single-AZ placement.

One Zone is appropriate when the workload can tolerate single-AZ storage placement, or when the data is replicated or backed up elsewhere for recovery.

EFS storage classes

EFS provides storage classes that balance access frequency, latency, and cost:

• EFS Standard: SSD-backed storage for frequently accessed data and low-latency requirements.
• EFS Infrequent Access (IA): Lower-cost storage for data that is accessed less frequently.
• EFS Archive: A lower-cost class for data that is rarely accessed, such as data accessed only occasionally for compliance or historical purposes.

All EFS storage classes provide an advertised durability target of 11 nines. Durability should not be confused with availability: a durable service can preserve data while still being temporarily unavailable.

Cross-Region replication

EFS can replicate a file system to another AWS Region for disaster recovery. The destination file system can be mounted in the target Region for read access, but it is read-only until a failover process promotes it for use as the active copy.

The destination mount target is not automatically available in every Availability Zone. Create the required mount target before attempting to mount the replicated file system from workloads in the destination Region.

EFS replication is intended for disaster recovery and provides recovery point and recovery time objectives measured in minutes, rather than zero-second replication or instant failover.

On-premises access

On-premises Linux clients can access EFS by using NFS over private network connectivity such as:

• AWS Site-to-Site VPN
• AWS Direct Connect

The on-premises environment must have appropriate routing, network reachability, and NFS-compatible clients. EFS is not a general-purpose Windows file share.

Backup integration

EFS integrates with AWS Backup for managed, scheduled backups. Use AWS Backup when the architecture requires centralized backup policies, retention, or cross-account and cross-Region backup governance.

Replication and backup solve different problems:

• Replication supports faster disaster recovery and maintaining a secondary copy.
• Backup supports point-in-time recovery, retention, and protection from accidental deletion or corruption.

Throughput modes

EFS performance can be selected independently from the amount of data stored:

• Provisioned throughput: Specify the desired throughput regardless of file system size. This is useful when a relatively small file system requires sustained throughput beyond what its stored capacity would naturally provide.
• Bursting throughput: Throughput scales with the amount of data stored, with the ability to burst to higher levels. Larger file systems generally receive more baseline throughput.

The correct choice depends on workload access patterns, required sustained throughput, file system size, and cost. Do not assume that a small file system automatically provides the same throughput as a large one when using bursting mode.

Exam-Relevant Takeaways

• EFS is a shared, elastic file system for Linux workloads using NFS.
• A regional EFS file system is the normal choice when the application requires multi-AZ storage resilience.
• Mount instances to the EFS mount target in the same Availability Zone whenever possible.
• One Zone EFS is single-AZ storage and should be selected only when the cost or placement tradeoff is acceptable.
• EFS replication creates a read-only destination until failover.
• EFS replication is asynchronous disaster recovery with RPO and RTO measured in minutes.
• A replicated destination still requires mount targets in the Availability Zones where clients will run.
• EFS IA and Archive reduce storage cost for less frequently accessed data, while Standard targets frequently accessed data.
• Provisioned throughput is useful when throughput requirements are high relative to the amount of data stored.
• AWS Backup can provide managed, policy-based EFS backups; replication alone is not a replacement for backup.

Architecture Decision Guide

Requirement	Recommended EFS design	Main consideration
Shared files across EC2 instances in multiple AZs	Regional EFS	Provides multi-AZ storage resilience and shared access
Lowest-cost shared file storage with single-AZ tolerance	One Zone EFS	Data is placed in one AZ; plan backup or replication
Frequently accessed, low-latency files	EFS Standard	Higher storage cost than infrequent-access classes
Files accessed periodically	EFS IA	Lower storage cost, with access-related tradeoffs
Long-term, rarely accessed files	EFS Archive	Lowest-cost tier for very infrequent access
DR copy in another Region	EFS replication	Destination is read-only until failover; RPO/RTO are measured in minutes
Centralized retention and recovery policies	AWS Backup for EFS	Configure backup plans, retention, and recovery procedures
Throughput must not depend on stored capacity	Provisioned throughput	Pay for explicitly configured throughput
Throughput can scale with file system size	Bursting throughput	Baseline throughput is related to stored data, with burst capability
Linux clients outside AWS	VPN or Direct Connect with NFS	Requires private routing and NFS-compatible clients

Common Exam Traps

• Choosing EFS for Windows clients: EFS is intended for Linux clients using NFS. A Windows SMB file-share requirement points toward a different service, such as Amazon FSx for Windows File Server.
• Confusing regional EFS with a single mount target: A regional file system uses mount targets associated with Availability Zones. Applications should connect through the local-AZ mount target.
• Treating One Zone as multi-AZ storage: A One Zone file system can be accessed cross-AZ, but its underlying data remains in one Availability Zone.
• Assuming replication is active-active: The replicated EFS destination is read-only until failover; it is not an independently writable active-active file system.
• Forgetting destination mount targets: A replicated file system does not automatically provide a mount target in every required AZ.
• Using replication as a backup strategy: Replication helps with regional disaster recovery but does not replace backups with retention and recovery-point requirements.
• Assuming throughput is always independent of file system size: That is true for provisioned throughput, not for bursting throughput.
• Confusing durability with availability: The 11-nines durability target does not mean the file system can never be unavailable.

Real-World Engineer Notes

• Use regional EFS for highly available application tiers that need a shared POSIX-style file system across Availability Zones.
• Place mount targets where workloads run and use local-AZ connections to reduce latency and avoid unnecessary inter-AZ data transfer.
• Treat EFS replication failover as an operational procedure. Test promotion, remounting, DNS or endpoint changes, and application recovery before a disaster occurs.
• For One Zone deployments, explicitly document the recovery source, such as AWS Backup or cross-Region replication.
• Select storage classes based on actual access patterns rather than simply choosing the lowest storage price. Retrieval behavior and application latency matter.
• Throughput planning should consider concurrent clients, file sizes, metadata operations, and sustained versus burst access patterns.
• Secure EFS through VPC networking, mount-target security groups, and appropriate identity and file-system permissions. Network reachability alone should not grant application access.

Quick Reference Summary

• Service: Amazon Elastic File System
• Primary use: Shared file storage for Linux workloads
• Protocol: NFS version 4
• Regional option: Multi-AZ resilient file system with mount targets in AZs
• One Zone option: Single-AZ file system with lower resilience and potentially lower cost
• Storage classes: Standard, Infrequent Access, and Archive
• Durability target: 11 nines across storage classes
• DR: Cross-Region replication with a read-only destination until failover
• Backup: AWS Backup integration
• On-premises connectivity: NFS over VPN or Direct Connect
• Throughput modes covered: Provisioned and bursting

Flashcards

1. Q: What protocol does Amazon EFS use?

A: NFS version 4.

1. Q: What operating system family is EFS designed to support?

A: Linux-based clients.

1. Q: What is the main characteristic of a regional EFS file system?

A: Its data is designed for multi-AZ resilience, with mount targets associated with Availability Zones.

1. Q: Where should an EC2 instance normally mount EFS from?

A: Through the mount target in the same Availability Zone.

1. Q: What is the main tradeoff of One Zone EFS?

A: Lower-cost single-AZ placement in exchange for reduced AZ-level resilience.

1. Q: Can a client in another AZ access a One Zone file system?

A: Yes, but the access is cross-AZ and the data remains in the original AZ.

1. Q: What is EFS Standard intended for?

A: Frequently accessed data requiring low-latency SSD-backed storage.

1. Q: When should EFS IA or Archive be considered?

A: When data is accessed less frequently and reducing storage cost is more important than frequent low-latency access.

1. Q: What is the state of an EFS replication destination before failover?

A: Read-only.

1. Q: What are typical EFS replication recovery objectives?

A: RPO and RTO measured in minutes.

1. Q: Does EFS replication automatically create every required mount target in the destination Region?

A: No. Create mount targets in the AZs where destination workloads need access.

1. Q: What is the purpose of provisioned throughput?

A: To specify throughput independently of the amount of data stored.

1. Q: How does bursting throughput generally relate to file system size?

A: Baseline throughput scales with the amount of data stored, with the ability to burst higher.

1. Q: What AWS service provides managed EFS backups?

A: AWS Backup.

Practice Questions

Question 1

A company runs a Linux application tier across three Availability Zones. All instances must access the same shared file system, and the file system must remain available if one Availability Zone fails. Which design best meets the requirement?

A. One Zone EFS with mount targets in all three Availability Zones
B. Regional EFS with mount targets for the required Availability Zones
C. EBS Multi-Attach volumes mounted by all instances
D. Amazon S3 mounted through an NFS gateway

Correct answer: B

A regional EFS file system is designed for shared access across Availability Zones and multi-AZ storage resilience. One Zone EFS remains single-AZ even if clients can access it cross-AZ.

Question 2

A workload stores 2 TB of Linux application files in EFS. The files are usually accessed several times each day and require low latency. Which storage class is the best initial choice?

A. EFS Archive
B. EFS Infrequent Access
C. EFS Standard
D. One Zone Archive only

Correct answer: C

EFS Standard is intended for frequently accessed data and low-latency performance. IA and Archive are more appropriate when access frequency is lower.

Question 3

A company replicates an EFS file system from Region A to Region B. During a regional outage, administrators want applications in Region B to use the replicated data. What must they account for?

A. The replica is always writable and active-active
B. The replica is read-only until failover, and required mount targets must exist in Region B
C. EFS replicas can only be accessed from Region A
D. Replication automatically converts EFS into an SMB file share

Correct answer: B

The destination is read-only before failover. The recovery Region also needs mount targets in the Availability Zones where applications will run.

Question 4

An organization has a small EFS file system but requires a predictable amount of sustained throughput that should not depend on storage size. Which throughput choice is most appropriate?

A. Bursting throughput
B. Provisioned throughput
C. Archive storage class only
D. Cross-Region replication

Correct answer: B

Provisioned throughput allows the architect to specify throughput independently of the amount of data stored. Bursting throughput is related to file system size.

Question 5

An on-premises Linux application needs to mount an EFS file system in a VPC. Which connectivity design is appropriate?

A. Public Internet access using SMB
B. VPN or Direct Connect with NFS and appropriate routing
C. AWS Storage Gateway cached volumes using iSCSI only
D. An S3 bucket policy without network connectivity

Correct answer: B

On-premises Linux clients can access EFS through NFS over private connectivity such as Site-to-Site VPN or Direct Connect. Routing and security controls must also permit the connection.