Data transfer and migration: DataSync, Snow Family, Transfer Family, DMS, Application Migration Service
8 min read · about 55 min with practice3 quick checks≈2% of the testCore: Core: tested on most papers
Reading is free. Sign in to tick off lessons, keep your place and track your mastery.
Migration questions span tasks 3.5, 4.1 and 4.3. They reward two skills: a quick transfer-time estimate that tells you whether the network is fast enough, and knowing which migration service moves which kind of thing. Files go with DataSync, file uploads from partners with Transfer Family, databases with DMS and whole servers with Application Migration Service. Most wrong options are real services pointed at the wrong kind of data.
By the end you’ll be able to
Calculate whether a dataset can move over the available bandwidth in time, and choose DataSync, Snowball Edge or Direct Connect accordingly
Select DataSync (scheduled, incremental, NFS/SMB/HDFS to S3/EFS/FSx) versus Storage Gateway (ongoing hybrid access)
Plan homogeneous and heterogeneous database migrations with DMS, DMS Schema Conversion/SCT and ongoing replication (CDC)
Rehost servers with AWS Application Migration Service and choose among the migration strategies for a given portfolio
What the exam asks
Online or offline? Estimate whether the data can cross the available bandwidth before the deadline. If it can, use DataSync, over Direct Connect where one exists. If it cannot, use the Snow Family.
One-time move or ongoing hybrid access? A move or a scheduled copy is DataSync. Continuing on-premises access to cloud storage is Storage Gateway.
Partner file exchange. SFTP, FTPS, FTP or AS2 into S3 or EFS is AWS Transfer Family.
Databases. Plan homogeneous and heterogeneous moves with AWS DMS, add DMS Schema Conversion or the Schema Conversion Tool when the engine changes, and use full load plus change data capture (CDC) for minimal downtime.
Servers and portfolios. Rehost with AWS Application Migration Service (MGN), and classify each application with the migration strategies (the 7 Rs).
Core ideas
Transfer-time estimate (do this in your head)
Bits to bytes: divide by 8. There are 86,400 seconds in a day.
Usable bandwidth
Throughput
Per day (approx.)
100 Mbps
12.5 MB/s
about 1 TB
500 Mbps
62.5 MB/s
about 5.4 TB
1 Gbps
125 MB/s
about 10.8 TB
10 Gbps
1.25 GB/s
about 108 TB
Days = data ÷ (TB per day × share of the link you can actually use). For example, 300 TB over half of a 500 Mbps link is 300 ÷ 2.7, or about 110 days. If the result is longer than the deadline, the network cannot carry the data. Remember that a new Direct Connect connection usually takes weeks to months to provision, so it is rarely the answer for an urgent one-time move.
AWS DataSync
What it is: a managed online transfer service. An agent, a VM deployed next to your storage, reads NFS, SMB, HDFS or self-managed object storage. DataSync also copies between AWS storage services and from other clouds.
vii.Check your understanding
3 questions on data transfer and migration: DataSync, Snow Family, Transfer Family, DMS, Application Migration Service. Every option is explained once you answer.
Sign in to try the quick check
Answers are checked on our side, every option is explained, and your result feeds your mastery for this topic. It’s free.
The first 3 of 10 cards for this topic. Sign in and finish the lesson to review them with spaced repetition.
PromptCard 1 of 3
What is the rule of thumb for transfer time?
Destinations: S3 (any storage class), EFS, and every FSx file system.
Features: scheduling, incremental copies of changed data only, in-line integrity verification, preservation of metadata and permissions, encryption in transit, bandwidth throttling and filters.
Private path: activate the agent with a DataSync interface VPC endpoint (PrivateLink) reached over a Direct Connect private VIF or a VPN. Traffic then never touches a public endpoint.
Snow Family (offline transfer)
A Snowball Edge Storage Optimized device is shipped to your site. You copy data onto it locally, ship it back, and AWS imports the data into S3. Encryption is enforced, and several devices can work in parallel. The exam guide still lists the Snow Family, and the exam expects it when “limited bandwidth” meets “large dataset” and “fixed deadline”.
Storage Gateway compared with DataSync
Need
Service
Move data once, or copy it on a schedule, then use it in AWS
DataSync
On-premises apps keep reading and writing via NFS or SMB, with the data stored in S3 and a local cache for low latency
S3 File Gateway
Block volumes backed by S3 snapshots, or a virtual tape library
Volume Gateway or Tape Gateway
AWS Transfer Family
A fully managed server for SFTP, FTPS, FTP and AS2 that stores files in S3 or EFS. Partners keep their existing clients.
Identity comes from service-managed users (SSH keys), AWS Directory Service or a custom identity provider (Lambda or API Gateway).
Endpoint types: a public endpoint uses AWS-managed IP addresses that can change. A VPC-hosted, internet-facing endpoint carries Elastic IP addresses that partners can allow-list, and security groups restrict which source IPs can connect. Plain FTP is supported only on a VPC endpoint.
AWS DMS
Homogeneous migrations keep the engine, for example MySQL to RDS for MySQL or SQL Server to RDS for SQL Server.
Heterogeneous migrations change the engine, for example Oracle to Aurora PostgreSQL. First convert the schema, views and stored procedures with DMS Schema Conversion or the AWS Schema Conversion Tool (SCT), then move the data with DMS.
Full load + CDC copies existing data, then streams ongoing changes so the target stays in sync while the source remains online. Cut over when replication lag is near zero. The outage lasts minutes.
DMS can also validate data, and it can run as DMS Serverless so there is no replication instance to size.
AWS Application Migration Service (MGN)
MGN is the rehost tool for physical, VMware, Hyper-V and other-cloud servers. An agent replicates each server’s disks continuously at block level into a staging area. You launch test instances, verify them, and cut over with minutes of downtime. It replaces CloudEndure Migration and Server Migration Service.
Migration strategies (7 Rs)
Strategy
Meaning
Example
Retire
Switch it off
Unused legacy reporting app
Retain
Keep on premises for now
Mainframe with a 3-year contract
Rehost
Lift and shift
VMware VMs to EC2 with MGN
Relocate
Move the hypervisor platform
VMware estate moved as-is to a VMware cloud
Replatform
Lift, tinker and shift
Self-managed MySQL to RDS for MySQL
Repurchase
Replace with SaaS
On-premises CRM replaced by a SaaS CRM
Refactor
Re-architect cloud-native
Monolith rebuilt on Lambda and DynamoDB
Worked examples
Exam technique
Do the arithmetic first. Compare data ÷ (TB per day × usable share) with the deadline. Ship devices only if the network loses.
Name the object being moved: files, server, database or partner uploads. That alone removes two options.
For large moves with live changes, combine offline and online. Snowball Edge carries the bulk copy, then DataSync copies the delta, then you cut over.
“Minimal downtime” plus a database means DMS full load and CDC. “Engine change” adds schema conversion.
“Continue to access from on premises” means Storage Gateway, not DataSync.
Common mistakes
Quick recap
Estimate first: 1 Gbps moves about 10.8 TB a day, and 100 Mbps about 1 TB a day. Multiply by the share of the link you can use.
DataSync handles online, scheduled, incremental, verified copies from NFS, SMB, HDFS or object storage to S3, EFS or FSx, privately through an interface endpoint.
The Snow Family handles offline bulk transfer when the network loses the race. It is still on the exam, but closed to new customers in practice.
Storage Gateway provides ongoing hybrid access. Transfer Family provides managed SFTP, FTPS, FTP and AS2 to S3 or EFS, with Elastic IPs on a VPC endpoint.
DMS with full load and CDC gives minimal-downtime database moves. Add Schema Conversion or SCT when the engine changes.
MGN rehosts servers with continuous replication, test launches and a cutover of minutes. Use the 7 Rs to classify the rest of the portfolio.