AWS Snowball User Guide
Planning Your Large Transfer
• Step 6: Separate Your Data into Transfer Segments (p. 38)
Step 1: Understand What You're Moving to the Cloud
Before you create your first job for Snowball, you should make sure that you know what data you want
to transfer, where it is currently stored, and the destination you want to transfer it to. For data transfers
that are a petabyte in scale or larger, doing this administrative housekeeping makes your life much easier
when your Snowballs start to arrive.
You can keep this data in a spreadsheet or on a whiteboard—however it works best for you to organize
the large amount of content you plan to move to the cloud. If you're migrating data into the cloud for
the first time, we recommend that you design a cloud migration model. For more information, see the
whitepaper A Practical Guide to Cloud Migration on the AWS Whitepapers website.
When you're done with this step, you know the total amount of data that you're going to move into the
cloud.
Step 2: Prepare Your Workstations
When you transfer data to a Snowball, you do so through the Snowball client, which is installed on a
physical workstation that hosts the data that you want to transfer. Because the workstation is considered
to be the bottleneck for transferring data, we highly recommend that your workstation be a powerful
computer, able to meet high demands in terms of processing, memory, and networking.
For large jobs, you might want to use multiple workstations. Make sure that your workstations all meet
the suggested specifications to reduce your total transfer time. For more information, see Workstation
Specifications (p. 93).
Step 3: Calculate Your Target Transfer Rate
It's important to estimate how quickly you can transfer data to the Snowballs connected to each of your
workstations. This estimated speed equals your target transfer rate. This rate is the rate at which you can
expect data to move into a Snowball given the realities of your local network architecture.
By reducing the hops between your workstation running the Snowball client and the Snowball, you
reduce the time it takes for each transfer. We recommend hosting the data that you want transferred
onto the Snowball on the workstation that you transfer the data through.
To calculate your target transfer rate, download the Snowball client and run the snowball test
command from the workstation that you transfer the data through. If you plan on using more than one
Snowball at a time, run this test from each workstation. For more information on running the test, see
Testing Your Data Transfer with the Snowball Client (p. 53).
While determining your target transfer speed, keep in mind that it is affected by a number of factors
including local network speed, file size, and the speed at which data can be read from your local servers.
The Snowball client copies data to the Snowball as fast as conditions allow. It can take as little as a
day to fill a 50 TB Snowball depending on your local environment. You can copy twice as much data in
the same amount of time by using two 50 TB Snowballs in parallel. Alternatively, you can fill an 80 TB
Snowball in two and a half days.
Step 4: Determine How Many Snowballs You Need
Using the total amount of data you're going to move into the cloud, found in step 1, determine how
many Snowballs you need to finish your large-scale data migration. Remember that Snowballs come in
50 TB (42 usable) and 80 TB (72 usable) sizes so that you can determine this number effectively. You can
move a petabyte of data in as little as 14 Snowballs.
37