Which Tool Migrates 50 TB With Least Overhead?

SAA-C03 By islever Editorial Last updated

Determine high-performing data ingestion and transformation solutions. Design cost-optimized compute solutions.

Short answer

C Order an AWS Snowball Edge Storage Optimized device to ship the 50 TB and run the transformation as a serverless AWS Glue custom job in the cloud.

Question

A company uses 50 TB of data for reporting. The company wants to move this data from on premises to AWS. A custom application in the company’s data center runs a weekly data transformation job. The company plans to pause the application until the data transfer is complete and needs to begin the transfer process as soon as possible.
The data center does not have any available network bandwidth for additional workloads. A solutions architect must transfer the data and must configure the transformation job to continue to run in the AWS Cloud.
Which solution will meet these requirements with the LEAST operational overhead?
  1. Use AWS DataSync to move the data. Create a custom transformation job by using AWS Glue.
  2. Order an AWS Snowcone device to move the data. Deploy the transformation application to the device.
  3. Order an AWS Snowball Edge Storage Optimized device. Copy the data to the device. Create a custom transformation job by using AWS Glue. Correct Answer
  4. Order an AWS Snowball Edge Storage Optimized device that includes Amazon EC2 compute. Copy the data to the device. Create a new EC2 instance on AWS to run the transformation application.

Key Takeaways

The question tests choosing a physical data-transfer service when no network bandwidth exists, plus a serverless transformation service; the trap is assuming EC2 or a device-resident application is lower effort than a managed Glue job.

This SAA-C03 scenario asks how to move 50 TB from a bandwidth-starved data center to AWS while keeping a weekly transformation job running in the cloud with minimal operational overhead. The correct path is an AWS Snowball Edge Storage Optimized device for the transfer plus an AWS Glue custom ETL job for the transformation, because Glue is serverless and eliminates instance management.

Common mistake: Option D is the most-chosen wrong answer: learners reason that re-implementing the custom job in Glue is heavy lifting, but they overlook that D provisions and operates a new EC2 instance (patching, scaling, monitoring) and that its Snowball Edge compute capacity is never even used.

Expert Analysis

Why the Answer Is Correct

Option C pairs an AWS Snowball Edge Storage Optimized device — rated for 80 TB usable, so it comfortably holds the 50 TB dataset — with AWS Glue for the weekly transformation, which meets the "no available network bandwidth" and "begin as soon as possible" constraints without saturating the WAN. Because the requirement is explicitly LEAST operational overhead, Glue's serverless model wins: no AMIs to bake, no EC2 patching, no autoscaling or instance-lifecycle management, and no cluster provisioning. Glue also runs custom ETL code (Spark/Python scripts), so the company's custom transformation logic can be expressed as a Glue job rather than as an always-managed application. Once the Snowball lands the data in Amazon S3, Glue reads directly from S3 and the weekly schedule continues in the cloud, fully satisfying the "continue to run in the AWS Cloud" clause.

Why the Other Options Are Wrong

Option A fails on the first constraint: AWS DataSync replicates over the network, and the data center "does not have any available network bandwidth for additional workloads," so a 50 TB DataSync task would take weeks and would compete with existing traffic. Option B fails on capacity: Snowball Edge Storage Optimized's smaller sibling, Snowcone, tops out around 14 TB SSD, far short of the required 50 TB, and it offers no meaningful way to run a transformation workload at this scale. Option D transfers the data correctly but then requires launching and maintaining a new Amazon EC2 instance to host the application — a server-based, patch-and-monitor model that carries strictly more ongoing operational overhead than Glue. Notably, D also wastes its premise: the option says to "Create a new EC2 instance on AWS," so the compute-capable Snowball Edge device it orders is never used for the job anyway.

Community Comment Notes

Commenters split between C and D, with 123jhl0 and studynoplay arguing for C on capacity and serverless grounds — "LEAST operational overhead = Serverless = Glue" — while goodmail and GB_12345 push D, GB_12345 asserting that "Snowball Edge doesn't support Glue". That objection misreads the design: Glue runs in the AWS Region after the Snowball ships data to S3, not on the device, so device service support is irrelevant. f761d0e objected that "The q states that the custom job must remain, so glue is out," but Glue supports custom ETL scripts, and the question asks for the job to continue to run in the cloud, not for identical code packaging. remand similarly warned that rebuilding in Glue is "reinventing the wheel," yet under the question's explicit "LEAST operational overhead" test, ongoing EC2 administration outweighs one-time job migration effort.

Official Reference

These links open the official vendor documentation in a new tab.

Exam Strategy

When a question forbids additional network bandwidth, immediately eliminate DataSync, Direct Connect-based, and S3-transfer-over-internet options, then check device capacity against the dataset size. On 'least operational overhead' phrasing, prefer serverless managed services (Glue, Lambda, Fargate) over EC2-based answers unless the workload must run unmodified on a specific runtime.

Frequently Asked Questions

Why can't AWS DataSync be used when the data center has no free bandwidth?

DataSync moves data over the network, so a 50 TB copy task would consume WAN capacity the site does not have and take weeks, violating the 'begin as soon as possible' requirement.

Doesn't running the transformation on EC2 mean less work than rebuilding it in AWS Glue?

EC2 means ongoing patching, monitoring, and instance lifecycle management. Glue is serverless and runs custom ETL scripts, so it carries lower operational overhead after the one-time job migration.

More SAA-C03 FAQ →

Discussion 72 contributions

Compiled from public community threads and reader submissions, lightly edited for clarity. Handles and vote tallies are shown as recorded, and comments reflect the views of individual contributors.

From public community threads 30 of 72

  1. 123jhl0 👍 77 Answered C
    A. Use AWS DataSync to move the data. Create a custom transformation job by using AWS Glue. - No BW available for DataSync, so "asap" will be weeks/months (?) B. Order an AWS Snowcone device to move the data. Deploy the transformation application to the device. - Snowcone will just store 14TB (SSD configuration). C. Order an AWS Snowball Edge Storage Optimized device. Copy the data to the device. Create a custom transformation job by using AWS Glue. - SnowBall can store 80TB (ok), takes around 1 week to move the device (faster than A), and AWS Glue allows to do ETL jobs. This is the answer. D. Order an AWS Snowball Edge Storage Optimized device that includes Amazon EC2 compute. Copy the data to the device. Create a new EC2 instance on AWS to run the transformation application. - Same as C, but the ETL job requires the deployment/configuration/maintenance of an EC2 instance, while Glue is serverless. This means D has more operational overhead than C.
  2. goodmail 👍 18 Answered D
    Why C? This answer misses the part between SnowBall and AWS Glue. D at least provides a full-step solution that copies data in snowball device, and installs the custom application in device's EC2 to do the transformation job.
  3. remand 👍 11
    I disagree on D. transformation job is already in place.so, all you have to do is deploy and run on ec2. C takes more effort to build Glue process, like reinventing the wheel . this is unnecessary
  4. GB_12345 👍 7 Answered D
    Not A. AWS DataSync requires an internet connection & the question states no available bandwidth Not B. SnowCone only has a max of 14 TB with an SSD, and the data is 50 TB Not C. Snowball Edge doesn't support Glue Supported services: https://docs.aws.amazon.com/snowball/latest/developer-guide/whatisedge.html So the answer must be D, as Snowball Edge Storage Optimized does support EC2 & can store 80 TB for the version that support compute resources
  5. jdr75 👍 6
    I agree. When it said "with least Operational overhead" , it does not takes in account "migration activities" neccesary to reach the "final photo/scenario". In "operational overhead" schema, you're situated in a "final scenario" and you've only take into account how do you operate it, and if the operation of that scheme is ALIGHTED (least effort to operate than original scenario), that's the desired state.
Show 25 more comments
  1. pentium75 👍 6
    Because AWS Glue means less "operational (!) overhead" than running an EC2 instance.
  2. cookieMr 👍 5
    Option A (AWS DataSync with AWS Glue) involves using AWS DataSync for data transfer, which requires available network bandwidth. Since the data center has no additional network bandwidth, this option is not suitable. Option B (AWS Snowcone device with deployment) is designed for smaller workloads and may not have enough storage capacity for transferring 50 TB of data. Additionally, deploying the transformation application on the Snowcone device could introduce complexity and operational overhead. Option D (AWS Snowball Edge with EC2 compute) involves transferring the data using a Snowball Edge device and then creating a new EC2 instance in AWS to run the transformation application. This option adds additional complexity and operational overhead of managing an EC2 instance. In comparison, option C offers a straightforward and efficient approach. The Snowball Edge Storage Optimized device can handle the large data transfer without relying on network bandwidth. Once the data is transferred, AWS Glue can be used to create the transformation job, ensuring the continuity of the application's processing in the AWS Cloud.
  3. Robrobtutu 👍 5
    Answer C uses Glue, answer D uses EC2, so I believe you probably meant you're voting for C.
  4. studynoplay 👍 5 Answered C
    LEAST operational overhead = Serverless = Glue
  5. Buruguduystunstugudunstuy 👍 4 Answered D
    The solution that will meet these requirements with the least operational overhead is Option D: (Order an AWS Snowball Edge Storage Optimized device that includes Amazon EC2 compute. Copy the data to the device. Create a new EC2 instance on AWS to run the transformation application.) AWS Snowball Edge Storage Optimized devices are used to transfer large amounts of data quickly and securely to and from the cloud. They come with onboard storage and compute capabilities, which allows you to perform data processing tasks on the device itself before transferring the data to the cloud. This means that you can copy the data to the device and then use the device's computing capabilities to run the transformation application directly on the device, without having to pause the application or transfer it to the cloud.
  6. G3 👍 4
    .... and the AI maven says : A solution that would meet these requirements with the least operational overhead is to use AWS Snowball Edge. Snowball Edge is a data transfer device that can transfer large amounts of data into and out of the AWS cloud with minimal network bandwidth requirements. Additionally, Snowball Edge can run custom scripts on the device, so the transformation job can be configured to continue running during the transfer. Once the transfer is complete, the data can be loaded into an AWS storage service such as Amazon S3. This solution would minimize operational overhead by allowing for a parallel transfer and processing of data, rather than requiring the application to be paused.
  7. small_zipgeniuis 👍 4 Answered C
    A - no bandwith, option out B - snowcone SSD has max 14TB of capacity C - is correct one here D - cannot use Compute optimized as max capactiy for this snowball is 39.5TB, and only that's why ;-) https://docs.aws.amazon.com/snowball/latest/developer-guide/device-differences.html
  8. f761d0e 👍 4
    The q states that the custom job must remain, so glue is out. Seems like D is the only option as DataSync needs bandwidth.
  9. NV305 👍 4 Answered C
    c only Glue is serverless. This means D has more operational overhead than C.
  10. PaulGa 👍 3 Answered D
    Ans D - but I suspect it should be Ans C because, as stated by 123jhlo (1yr 11mth ago), D is not serverless: "C. Order an AWS Snowball Edge Storage Optimized device. Copy the data to the device. Create a custom transformation job by using AWS Glue. - SnowBall can store 80TB (ok), takes around 1 week to move the device (faster than A), and AWS Glue allows to do ETL jobs." D. Order an AWS Snowball Edge Storage Optimized device that includes Amazon EC2 compute. Copy the data to the device. Create a new EC2 instance on AWS to run the transformation application. - Same as C, but the ETL job requires the deployment/configuration/maintenance of an EC2 instance, while Glue is serverless. This means D has more operational overhead than C."
  11. bignatov 👍 3 Answered D
    A and D are wrong. D is the correct answer. Why not C: AWS Snowball Edge is a physical device designed for transferring large amounts of data to and from AWS. It includes some compute capabilities, such as running AWS Lambda functions, AWS IoT Greengrass, and EC2 instances, but it does not support AWS Glue. AWS Glue is a fully managed ETL (Extract, Transform, Load) service that runs within the AWS Cloud. It is designed to work with data stored in AWS services like Amazon S3, Amazon RDS, and Amazon Redshift, among others. AWS Glue is not available on edge devices like Snowball Edge.
  12. Femmyte 👍 3 Answered D
    The answer is D because of the following key points 1. A custom application in the company’s data center runs a weekly data transformation job. Which means that the company already has an application that runs the transformation. 2. A solutions architect must transfer the data and must configure the transformation job to continue to run in the AWS Cloud. This shows that the only responsibility of the architect is to transfer the data and configure the existing application to run on the EC2 the architect is going to deploy.
  13. architect_kags 👍 3 Answered D
    Since the transformation application already there, Snowball Edge devices with compute capability allow to run applications directly on the device. This means the transformation job can start locally on the Snowball Edge device if needed or directly after the data transfer to AWS, minimizing delays. AWS Glue for transformation adds additional refactoring work and operational overhead.
  14. Robrobtutu 👍 3 Answered C
    I'm voting C and not D because creating a new EC2 instance in Snowball to run the transformation application has more overhead than running Glue. Another thing to consider is that answer C does not mandate us to install Glue in Snowball, we can run Glue after the data has been uploaded from Snowball to AWS.
  15. StuMoz 👍 3
    I was originally going to vote for C, however it is D because of 2 reasons. 1) AWS love to promote their own products, so Glue is most likely and 2) because Glue presents the least operational overhead moving forward as it is serverless unlike an EC2 instance which requires patching, feeding and watering
  16. Robrobtutu 👍 3
    The answer does not say you have to run Glue inside snowball edge, it just says you would use Glue, which could be after snowball edge reaches Amazon facilities and the data is uploaded to AWS.
  17. happpieee 👍 3
    AWS Glue is not part of SnowBall Edge AWS services it can run within. Check it out here : https://docs.aws.amazon.com/snowball/latest/developer-guide/whatisedge.html
  18. rcarmin 👍 3
    D answer says Snowball Edge STORAGE Optimized, which supports 80TB. 39.5TB is for the Snowball Edge COMPUTE Optimized.
  19. ocbn3wby 👍 3
    You can search this online. https://docs.aws.amazon.com/snowball/latest/developer-guide/device-differences.html
  20. Bang3R 👍 3 Answered C
    C has less operational overhead than D. Managing EC2 has higher operational overhead than serverless AWS Glue
  21. aba2s 👍 2 Answered C
    Option B is incorrect. Although you can use AWS DataSync to automate and accelerate data transfer from on-premises to AWS storage services, it’s not capable of replicating existing applications running on your server. Option B is incorrect as AWS Snowcone supports data collection and data processing using AWS compute services but supports only 8 TB of HDD-based hard disk. It’s not the best option for transferring 50 TB of data, as it will require multiple iterations of offline data transfer. I will go for C as it seem to have less operational overhead.
  22. ocbn3wby 👍 2 Answered D
    I would stick to D answer. In real life - this is what it would have happened. Maybe after the migration, the existing ETL application would be refactored to AWS services. But this takes development time vs "lift and shift" approach. Edge Storage optimized offers EC2 compute functionality (with AMI directly integrated) https://docs.aws.amazon.com/snowball/latest/developer-guide/device-differences.html
  23. live_reply_developers 👍 2 Answered D
    "A custom application in the company’s data center runs a weekly data transformation job." "A solutions architect must transfer the data and must configure the transformation job to continue to run in the AWS Cloud." LEAST operational overhead -> just take app and put on EC2, instead of configuring Glue
  24. tom_cruise 👍 2 Answered D
    The only way C can work is by using this: Amazon S3 adapter — Use for programmatic data transfer in to and out of AWS using the Amazon S3 API for Snowball Edge, but it is not indicated in the answer. If it includes the step to transfer the data from the Snowball to S3, then C would be a better answer.
  25. DavidNamy 👍 2 Answered D
    Option D, ordering an AWS Snowball Edge Storage Optimized device that includes Amazon EC2 compute, is the most efficient solution because it allows you to both transfer the data and run the transformation application on the same device, reducing the operational overhead required.

Reader comments & corrections

No reader corrections yet — spotted an error or have a note? Share it below.

Log in to comment, report an error, or add a note about this question.

Scheduled scaling for a predictable morning EC2 traffic spike

S-grade

Improving RDS Performance with Provisioned IOPS SSD

S-grade

S3 Multi-Region Access Point and Replication Strategy

S-grade

Securing NLB with Security Groups for Minimal Change

S-grade

Practice All SAA-C03 Questions

Access 120 questions with complete answers and detailed explanations.

View Full SAA-C03 Practice Test →

← Back to SAA-C03 Study Guide