SOA-C03 - Monitoring, Logging, Analysis, Remediation, and Performance Optimization (22% of the exam) - Section 1.6

Improve Amazon S3 and shared file storage performance using S3 Transfer Acceleration, multipart upload, AWS DataSync, Amazon EFS and Amazon FSx.

Improve Amazon S3 throughput using S3 Transfer Acceleration and multipart upload, transfer data to and from on-premises storage with AWS DataSync, and control costs on Amazon EFS through lifecycle policies. Recognise which technique addresses latency versus throughput versus cost for each storage tier.

S3 Transfer AccelerationMultipart uploadAWS DataSyncAmazon EFS lifecycle

Practice question for this objective

Free sampleMonitoring, Logging, Analysis, Remediation, and Performance Optimizationmedium

A company is decommissioning an on-premises Windows file server whose SMB shares hold several terabytes of files that depend on Windows ACLs. The data must land in an Amazon FSx for Windows File Server file system, the transfer must run as a managed scheduled task that verifies integrity, and the existing file permissions must be preserved on arrival. Which approach meets all of these needs with the least custom scripting?

AMount the FSx file system from a jump host and run robocopy with the copy-all flag on a Windows scheduled task to mirror the shares and their security descriptors.
BOrder an AWS Snowball Edge device, copy the shares onto it offline, and ship it so the data is imported directly into the FSx for Windows file system on return.
CEnable cross-Region replication from the on-premises SMB shares to FSx so that changed files and their permissions are continuously copied during the migration.
DDeploy an AWS DataSync agent on premises and configure a scheduled task from the SMB source to the FSx location that verifies data and preserves SMB metadata. Correct

Use AWS DataSync over SMB to migrate Windows file shares into Amazon FSx for Windows File Server with scheduling, verification and preserved permissions. AWS DataSync moves data between SMB file shares and Amazon FSx for Windows File Server through an on-premises agent. It runs as a scheduled task, transfers only changed files on each run, verifies that source and destination match, and copies SMB metadata such as ownership and ACLs so the Windows permission model arrives intact. This delivers a managed, repeatable migration without writing or maintaining custom copy scripts.

Why A is wrong: Robocopy can copy ACLs but is a hand-rolled script on a self-managed host with no built-in transfer verification, which is exactly the manual approach the requirement rules out.

Why B is wrong: Snowball suits one-off offline bulk movement but does not import directly into FSx or provide the scheduled, repeatable, integrity-verified sync the cutover requires.

Why C is wrong: Cross-Region replication is an S3 feature between buckets and cannot connect an on-premises SMB share to FSx, so it cannot perform this Windows file migration at all.

Why D is correct: DataSync supports SMB sources and FSx for Windows destinations, runs scheduled incremental tasks, verifies transferred data and carries the SMB metadata including ACLs, meeting every part of the requirement.

See more SOA-C03 practice questions, answers explained.

Exam traps in Monitoring, Logging, Analysis, Remediation, and Performance Optimization

Answers that look right on this material and are not. Each one is a distractor from a different question in the SOA-C03 bank for this domain.

Run repeated parallel rsync jobs from the NFS server to a mounted EFS target and schedule them with cron to keep the file system current.
Why it is wrong: A scripted rsync approach can copy files but is unmanaged, needs its own tuning and integrity checks, and is exactly the hand-rolled option the team wants to avoid.
Switch the file system to Provisioned throughput mode and set a fixed level high enough to cover the largest burst the workload has ever generated.
Why it is wrong: Provisioned throughput fixes a level the team must estimate and pay for continuously even while idle, which contradicts the wish to avoid sizing and pay only for usage.
Enable S3 Versioning on the bucket so that each interrupted attempt is retained as a separate version and the client can continue from the most recent stored version.
Why it is wrong: Versioning keeps multiple copies of completed objects for recovery, but it does nothing for in-flight upload speed and cannot resume a partial transfer, so it misses both stated needs.

Examworthy is not affiliated with or endorsed by Amazon Web Services. Original, blueprint-aligned practice material only.