Study guide
Technical reference and lesson notes
Purpose of This Lesson
The AWS Snow Family provides physical devices for transferring large data volumes to or from AWS when network-based transfer would take too long, be too expensive, or be impractical. Some Snow Family devices also provide compute and storage at the edge, allowing data to be collected and processed near its source before being sent to AWS.
The primary destination for transferred data is often an Amazon S3 data lake, although Snowball Edge can also support database migration workflows and other processing tasks.
Key Concepts
When to choose AWS Snow Family
Use a Snow Family device when:
- The dataset is very large.
- Available network bandwidth cannot complete the transfer within the required timeframe.
- The source location has limited or unreliable connectivity.
- Data must be collected or processed at a remote or disconnected site.
- Shipping encrypted physical storage is more practical than transferring over the network.
The overall duration includes data preparation, device delivery, loading, return shipping, and AWS ingestion. A network transfer may be preferable for smaller datasets or locations with sufficient bandwidth.
Snowball Edge
Snowball Edge devices combine physical data transfer with local storage and, depending on the configuration, compute capabilities.
Two important configurations are:
- Compute Optimized: Designed for workloads that need local processing before data is transferred to AWS. Multiple devices can be clustered for additional capacity and resilience.
- Storage Optimized: Designed to maximize the amount of data that can be moved per device. The source material identifies up to 210 TB of usable storage for this configuration.
Snowball Edge can write data to Amazon S3 after the device is returned to AWS. Objects retain their original or native data format, making the service suitable for bulk ingestion into an S3-based data lake.
Snowcone
Snowcone is a smaller and more portable Snow Family device. The source material identifies up to 14 TB of storage.
Its typical use case is edge data collection and processing, including data generated by:
- Sensors
- Internet of Things devices
- Industrial machinery
- Remote or mobile sites
Snowcone is useful when data must be handled locally first and then transferred online to AWS. AWS DataSync can be used to move data from Snowcone to AWS when network connectivity is available.
Security and encryption
Snow Family devices protect transferred data using AES-256 encryption. Data remains encrypted while stored on the device and during the transfer workflow. Device access and the associated job are managed through AWS controls, and the device is returned to AWS for processing after the customer loads the data.
For exam questions, distinguish physical device encryption from general network encryption. Snow Family is specifically intended to secure data while it is transported and stored on portable hardware.
Hadoop and database migration workflows
Snowball supports migration from Hadoop Distributed File System environments into Amazon S3. This allows data to be moved in its native format rather than requiring an intermediate conversion process.
Snowball Edge can also participate in database migration workflows with AWS Schema Conversion Tool (AWS SCT). In this pattern, data is staged on the device and loaded into the target database after the device arrives at AWS. AWS SCT addresses schema and code conversion concerns; the Snow device addresses the physical movement of the data.
Exam-Relevant Takeaways
- Choose AWS Snow Family when network transfer is too slow or impractical for the required data volume.
- Choose Snowball Edge Storage Optimized when the primary requirement is moving a large amount of data per device.
- Choose Snowball Edge Compute Optimized when data must be processed locally before transfer or when edge compute capacity is required.
- Choose Snowcone for smaller-scale, portable edge collection and processing.
- Use AWS DataSync when Snowcone or another edge location can transfer data online to AWS.
- Use Snow Family devices to populate Amazon S3 when shipping physical storage is faster than uploading over the network.
- Snow Family devices use AES-256 encryption for stored data.
- Snow Family is not automatically the best answer for every migration. Compare the expected network transfer time, shipping time, operational effort, and data security requirements.
- AWS SCT can support schema conversion in a database migration, but it does not replace the physical data-transfer capability of Snowball Edge.
Architecture Decision Guide
| Requirement | Recommended approach | Reason |
|---|---|---|
| Transfer a very large dataset from an on-premises site with limited bandwidth | Snowball Edge | Physical shipment can be faster than a network upload |
| Maximize storage capacity per shipped device | Snowball Edge Storage Optimized | Prioritizes data capacity |
| Process or filter data at a disconnected location before upload | Snowball Edge Compute Optimized | Provides local compute resources |
| Collect data from sensors at a remote or mobile site | Snowcone | Portable edge storage and compute |
| Move Snowcone data to AWS over an available network connection | AWS DataSync | Performs managed online data transfer |
| Load Hadoop data into an S3 data lake in its native format | Snowball with Hadoop-compatible workflow | Supports direct bulk movement into S3 |
| Perform schema conversion while moving database data | Snowball Edge with AWS SCT | Combines physical transfer with schema and code conversion |
| Transfer a moderate dataset over a fast, reliable connection | Online transfer services such as DataSync or direct upload | Avoids device logistics when the network is sufficient |
Common Exam Traps
- Confusing Snowball Edge with Snowcone: Snowcone is the smaller, more portable option. Snowball Edge is intended for larger transfers and can offer substantially greater storage and compute capacity.
- Assuming all Snow devices are only storage appliances: Snowball Edge configurations can provide local compute for edge processing.
- Selecting Snow Family solely because the dataset is large: The correct decision depends on bandwidth, transfer deadline, shipping time, connectivity, and operational constraints.
- Confusing DataSync with Snowball: DataSync transfers data online. Snowball physically transports encrypted storage devices.
- Assuming Snow Family transforms data automatically: Data is delivered to S3 in its native format. Transformation or analytics processing must be implemented separately.
- Treating AWS SCT as a transfer service: AWS SCT converts database schemas and code. It does not independently solve large-scale physical data movement.
- Ignoring return logistics: A Snowball workflow includes device shipping in both directions and AWS processing after return.
- Overlooking edge processing requirements: If the requirement is to filter, analyze, or act on data before it reaches AWS, a compute-capable Snow device may be more appropriate than a storage-only transfer method.
Real-World Engineer Notes
- Estimate the complete transfer timeline, not just the time required to copy data onto the device. Include ordering, provisioning, local copying, shipping, AWS ingestion, and validation.
- Calculate whether a network transfer is competitive. A high-bandwidth connection may be simpler and faster for smaller datasets, while a constrained link can make physical shipment more efficient.
- Validate application compatibility before selecting a compute-optimized device. Edge processing may require containerized workloads, local storage planning, and a strategy for handling intermittent connectivity.
- Encrypt data before and during transport, and apply least-privilege access to the AWS job and destination S3 bucket.
- Plan for integrity validation after import. Record object counts, checksums, manifests, and expected directory or partition structures.
- For an S3 data lake, define the destination bucket, prefix, lifecycle policies, and encryption configuration before the device arrives.
- Separate ingestion from downstream processing. Snow Family moves or stages the data; services such as AWS Glue, Amazon EMR, or analytics platforms can process it after ingestion when appropriate.
Quick Reference Summary
- Primary purpose: Offline or assisted bulk data transfer and edge computing.
- Common destination: Amazon S3.
- Snowball Edge Storage Optimized: High-capacity data transfer.
- Snowball Edge Compute Optimized: Local processing and compute-intensive edge workloads.
- Snowcone: Smaller portable device for edge collection and processing.
- Online transfer from edge: AWS DataSync.
- Database migration integration: AWS SCT can support schema conversion with Snowball Edge workflows.
- Encryption: AES-256.
- Key decision: Compare network transfer time and cost with device handling and shipping time.
Flashcards
- Q: When is AWS Snow Family preferred over an online upload?
A: When the data volume, bandwidth limitation, or connectivity constraints make a network transfer too slow or impractical.
- Q: What is the primary AWS destination for Snowball data in a data-lake migration?
A: Amazon S3.
- Q: Which Snowball Edge configuration is optimized for maximum storage capacity?
A: Storage Optimized.
- Q: Which Snowball Edge configuration is suitable for local processing before data transfer?
A: Compute Optimized.
- Q: What is Snowcone primarily designed for?
A: Portable, smaller-scale edge data collection and processing.
- Q: Which service can transfer Snowcone data online to AWS?
A: AWS DataSync.
- Q: What encryption algorithm protects data on Snow Family devices?
A: AES-256.
- Q: In what format does Snowball transfer data into S3?
A: The original or native data format.
- Q: How can Hadoop data be migrated into S3 using Snowball?
A: Through a Hadoop Distributed File System-compatible workflow that copies the data directly into S3 in its native format.
- Q: What does AWS SCT contribute to a Snowball-based database migration?
A: Schema and code conversion for the target database; Snowball provides the physical data-transfer mechanism.
- Q: What factors determine the total Snowball migration duration?
A: Data preparation and copy time, device shipping, return shipping, and AWS ingestion.
- Q: Why might Snowball Edge Compute Optimized be used at an edge site?
A: To analyze, filter, or otherwise process data locally before sending the resulting data to AWS.
Practice Questions
Question 1
A company must move 500 TB of archived files to Amazon S3. Its internet connection is heavily utilized, and the upload would take several weeks, but the migration must be completed within days. Which solution is most appropriate?
- A. Create an S3 bucket and upload the files over the existing connection
- B. Use AWS DataSync over the existing connection
- C. Order an appropriately sized Snowball Edge device, copy the data, and return it to AWS
- D. Use Amazon Kinesis Data Firehose to stream the archived files
Correct answer: C
Explanation: Snowball Edge is designed for large offline data transfers when network bandwidth cannot meet the required timeline. DataSync still relies on the network, and streaming services are not intended for bulk movement of a large archive.
Question 2
A mining site collects sensor data in a location with intermittent connectivity. The company wants to analyze and filter the data locally, then send selected results to AWS whenever connectivity is restored. Which option best meets the requirement?
- A. Snowball Edge Compute Optimized
- B. Snowball Edge Storage Optimized without local processing
- C. Amazon S3 Transfer Acceleration only
- D. AWS Direct Connect only
Correct answer: A
Explanation: Snowball Edge Compute Optimized provides local compute for processing data at the edge. The device can retain data during connectivity interruptions and transfer it when a connection is available.
Question 3
A remote facility needs a compact device to collect data from industrial equipment and later transfer the data online to AWS. The facility does not require the capacity or compute resources of a Snowball Edge. Which solution should be selected?
- A. Snowcone with AWS DataSync
- B. Snowball Edge Storage Optimized only
- C. AWS SCT
- D. Amazon S3 Glacier upload directly from the equipment
Correct answer: A
Explanation: Snowcone is the smaller portable Snow Family device for edge collection and processing. AWS DataSync can transfer its data online when connectivity is available.
Question 4
During a database migration, a company needs to physically move a large dataset and convert the source database schema for the target engine. Which combination is appropriate?
- A. Snowball Edge and AWS SCT
- B. Snowcone and Amazon CloudFront
- C. DataSync and Amazon Route 53
- D. S3 Transfer Acceleration and AWS WAF
Correct answer: A
Explanation: Snowball Edge handles large-scale physical data transfer, while AWS SCT helps convert database schemas and code for the target engine.
Question 5
An architect recommends Snowball solely because a dataset is large. Which additional analysis is most important before finalizing the design?
- A. Whether the S3 bucket can store any objects larger than 5 GB
- B. Whether network transfer time, shipping time, device handling, and security requirements make Snowball preferable
- C. Whether Snowball automatically transforms all files into columnar format
- D. Whether AWS SCT is required for every S3 migration
Correct answer: B
Explanation: Snowball is a tradeoff between online transfer and physical logistics. The architect must compare total elapsed time, operational effort, cost, connectivity, and security. Snowball does not automatically transform files, and AWS SCT is relevant to database schema conversion rather than every S3 migration.