AWS Snow Family

Overview

AWS Snowball is a highly secure, portable device that lets you collect and process data at the edge and migrate data in and out of AWS. For migrations of petabytes of data, Snowball may be a good use case.

There are two Snowball Edge device types, differing in storage:

Data Migration

Transferring data over a network takes time. Example: 100 TB over a 1 Gbit/s connection takes about 12 days.

Network transfer is a poor fit when you have:

• limited connectivity or limited bandwidth
• very high network cost
• shared bandwidth with other applications
• an unstable connection
• a transfer that would take over a week

Upload directly to S3 is simple but may use all your bandwidth. With Snowball:

1. You receive a physical Snowball device at your own infrastructure.
2. You load it with your data.
3. You ship it back to AWS.
4. AWS runs an export process that moves the data from the Snowball into, for example, an S3 bucket.

Edge Computing

Process data while it is created at an edge location, for example a truck on the road, a ship at sea or a mining station underground. These places may have no internet access or limited internet access and compute power.

• Order a Snowball Edge device and do edge computing on it. The Compute Optimized device is dedicated to this use case.
• You can run EC2 instances or Lambda functions directly on the device.
• Use cases: pre-process data, machine learning at the edge, transcode media at the edge.
• Once data is created and processed, send it back to AWS.

Hands-On Walkthrough

The demo goes through the ordering process without ordering a device.

1. Enter a job name (e.g. import job).
2. Choose the job type:
   • Import into Amazon S3: AWS ships a Snowball to you, you send it back.
   • Export from Amazon S3: data is loaded onto the Snowball and shipped to you.
   • Local compute and storage only: edge computing.
3. Choose the Snow device: Snowball Edge Storage Optimized (210 TB) or the Compute Optimized device. Many older devices have been discontinued and options may change; two options are enough for the exam.
4. Choose the pricing option (on-demand per-day pricing), the S3 data transfer and the target buckets.
5. Security: choose how to encrypt data and a service role. The role has access to S3 and lets AWS create a service-linked role so the Snow device can write to your S3 buckets.
6. Enter the shipping address and shipping speed (one-day or two-day).
7. Set notifications for the job status.
8. Review the job summary.

You receive a Snowball Edge, load your data, and send it back with the included shipping label.

Architecture: Snowball into Glacier

Solution:

1. Snowball imports the data into Amazon S3.
2. An S3 lifecycle policy transitions the objects into Amazon Glacier.

See S3 Lifecycle Rules.

Questions & Answers