AWS Comparisons

Service comparisons that drive exam decisions.

Use these pages when two or three services seem plausible and the deciding clue is operational overhead, security, scale, latency, or cost.

S3 vs EBS vs EFS vs FSx

These all store data, but they expose very different access models.

The main split is relational SQL versus key-value/document access patterns.

All decouple systems, but queueing, fanout, and event routing are not the same.

These can all front applications, but layer and feature set drive the answer.

Metrics/logs, API audit, and configuration history answer different questions.

Users are long-lived identities. Roles are assumed for temporary credentials.

Both can grant access, but they attach to different sides of the authorization decision.

Security groups protect resources. NACLs filter subnet traffic.

High availability and read scaling are different goals.

Backups help restore. Replication keeps another copy synchronized for availability or DR.

CloudFront is an HTTP CDN. Global Accelerator improves global network paths to application endpoints.

Both store configuration values, but rotation and secret lifecycle are the split.

Streams gives custom consumers and replay. Firehose gives managed delivery.

Both can process big data, but operations and control differ.

All run compute, but control and operational overhead vary.

Workflow state and message routing solve different problems.

Serverless S3 SQL and managed warehouse analytics are not the same workload.

Foundation model app, custom ML lifecycle, and prebuilt AI APIs are different levels of abstraction.

Improve model behavior by changing instructions, grounding context, or model weights.

The inference pattern follows latency and traffic shape.