AWS Solutions Architect Handbook SAA-C03, from zero Bipin Singh
Storage

Hybrid storage and data transfer

2 min readChapter 25 of 48By Bipin Singh

Many companies keep data on-premises while using the cloud, or need to move very large amounts of data into AWS. These services bridge that gap. Task statements 3.1, 3.5 and 4.1 all touch them.

AWS Storage Gateway

A hybrid service that gives on-premises applications access to cloud storage through standard protocols, with local caching.

Gateway type Protocol Backed by Use
S3 File Gateway NFS / SMB S3 (objects map to files) File shares backed by S3; data lands in S3 for analytics, lifecycle, backup
FSx File Gateway SMB FSx for Windows File Server Low-latency local access to FSx shares (AWS has limited availability of this type for new customers — check current status)
Volume Gateway — cached iSCSI S3, with frequently used data cached locally Primary data in AWS, low-latency local cache
Volume Gateway — stored iSCSI Full dataset local, asynchronously backed up to S3 as EBS snapshots Primary data on-premises, backups in AWS
Tape Gateway iSCSI virtual tape library S3 / Glacier archive tiers Replace physical tape backups without changing backup software
Key idea

"On-premises apps need file access to data that should live in S3" → S3 File Gateway. "Replace physical tapes, keep existing backup software" → Tape Gateway.

AWS DataSync

Online data transfer service that moves data between on-premises storage (NFS, SMB, HDFS, object storage) and AWS storage (S3, EFS, FSx), and between AWS storage services.

AWS Transfer Family

Fully managed SFTP, FTPS, FTP and AS2 endpoints that store files in S3 or EFS. Keep partners' existing file-transfer workflows unchanged while the data lands in AWS. Cue: "partners upload files via SFTP".

The AWS Snow Family (offline transfer and edge computing)

Physical, rugged devices that AWS ships to you:

Watch out

AWS has been narrowing the Snow Family over time (some devices discontinued or restricted to existing customers). Check current availability — but know the concept for the exam: physical device → offline transfer of tens of TB to PB, or compute at disconnected edge sites.

Choosing a transfer method

Rough transfer time over a network:

time ≈ data size ÷ effective bandwidth
100 TB over a 1 Gbps link at ~80% utilisation ≈ 100 × 8,000 Gb ÷ 0.8 Gbps ≈ 1,000,000 s ≈ 11.6 days
Situation Best fit
One-off or recurring file migration over the network DataSync
Ongoing hybrid access to cloud storage from on-premises apps Storage Gateway
Partners using SFTP/FTPS Transfer Family
Huge dataset, slow/limited network, deadline Snow Family device(s) (offline)
Fast uploads to S3 from far away S3 Transfer Acceleration
Consistent, high-bandwidth, ongoing transfer Direct Connect (+ DataSync over it)
Database migration AWS DMS (see migration)

Exam patterns

Bipin Singh
Written by Bipin Singh

Senior Full-Stack Engineer · AI & AWS. I design and run production systems on AWS — serverless, data and AI.

Work with me