DEA-C01 sample exam questions

AWS DMS migrates databases to RDS with minimal downtime, and it can keep replicating changes. Lambda would need custom code, Direct Connect is only a link, and DataSync moves files rather than database transactions.

The stream belongs in the security account, an IAM role lets CloudWatch Logs put data into it, and the subscription filter is created in the production account where the logs are. A filter in the security account would have no logs to read, and a stream in production does not deliver the data to the security account.

An existing gp2 volume can be changed to gp3 in place through Elastic Volumes, with no detachment or interruption. Snapshots, new volumes and data transfers all take more work and can cause disruption.

A separately attached EBS volume persists by default after an instance terminates, because delete on termination is off for added volumes. A root EBS volume is deleted by default, and instance store data is lost on termination.

A longer retention period keeps messages through downtime, and faster processing clears a backlog before messages expire. A dead-letter queue receives messages that fail processing, not those that simply expire. A longer visibility timeout and a delay queue do not prevent expiry.

A compound sort key works best when queries filter on its leading columns, so filtering on both Region ID and Department ID helps most, and the order in the WHERE clause does not matter. Filtering only on Region ID uses less of the key, and skipping Department ID or Region ID loses most of the benefit.

AWS Direct Connect avoids the public internet, and Secrets Manager securely stores service account credentials. Public subnets do not avoid the internet, and presigned URLs in S3 are unnecessary and do not manage these credentials.

A single EMR security configuration can hold both the KMS key for S3 at-rest encryption and the PEM file path for in-transit encryption, and it is specified when the cluster is created. Only one security configuration can apply to a cluster, it cannot be attached after creation, and local disk encryption does not cover S3.

A level-aware aggregate calculation computes at a chosen level, such as Region, before other aggregations, so the total does not change with drill-down. Table calculations and simple fields follow the visual, and LAC-W is for window functions over the data.

This is the only complete query that filters to rows with an error code and to 2024 onward, with a limit of 10. It orders by count, so it returns the most frequent errors rather than strictly the most recent, but the other options are less suitable.

AWS SCT converts a Teradata schema for Redshift, and AWS DMS can then help move the data. DMS Schema Conversion has more limited coverage of this source, automatic conversion is not available for this path, and manual schema work takes longer.

A scheduled Glue crawler adds new partitions and picks up schema changes. Partition projection needs a fixed schema configuration and does not detect changes, DataBrew is for preparation, and Lambda needs custom code.

Automatic compaction based on file size and file count thresholds keeps small files under control without a fixed schedule. A daily Glue job leaves many small files all day, compaction cannot be set to run every minute, and partitions and crawlers do not fix small files.

Domain units per business unit give isolated control, IAM Identity Center centralizes sign-in, and access requests allow approved sharing. API keys do not provide centralized authentication, restricting projects to owners prevents sharing, and separate domains make sharing and administration harder.

Step Functions coordinates Glue jobs on a schedule and on events, and OpenSearch Service provides full-text search of legal documents. RDS is not designed for complex full-text search, a custom Lambda search takes much more work, and scheduling alone does not support the 5-minute update.