在不同 AWS 账户的服务器之间共享文件

Question

I need to move several GB of data from one AWS account to another. The data moved will be loaded in an existing application. The storage shared between the two accounts will be temporary.

I need a simple solution and not necessarily the "best". I was thinking to use an S3 bucket, but then I thought that it would complicate the implementation of the migration because I would need to read from an S3 bucket instead of a simple file system.

What is an efficient and simple way to move many GB of data from a file system in one AWS account to a file system in a different AWS account?

Answer 1

You have some options, everything depends of the size and the locations of your data.

A specific service for this kind of situation is the AWS DataSync service, that you can communicate the accounts using VPC peering, but if you cannot peer VPCs together, perhaps due to internal security policies or regulatory compliance, you can still use the inte.net to securely transfer the data.

The DataSync service supports the following Locations (sources and destinations):

• Network File System (NFS)
• Server Message Block (SMB)
• Self-managed object storage
• Amazon EFS
• Amazon FSx for Windows File Server
• Amazon S3

Your architecture with DataSync can be like the following one:

More information about using Datasync can be found here Transferring data between AWS accounts using AWS DataSync

About the Pricing the AWS DataSync can be costly depending your locations, AWS charges the destination account for the use of AWS DataSync, since this is where you use the DataSync endpoint. Refer to the DataSync pricing page for more information.

Also, we have the following options that can be an option for you:

• Cross-Account Copying of EBS Snapshots
• Cross-account bulk transfer of files using Amazon S3 Batch Operations
• Copy data from an S3 bucket in one account and Region to another account and Region

Answer 2

Make use of Snapshot here. Share that with the another account > Create the file system their

This would be faster to implement also.