CertifyDeckCargando contenido…
DEA-C01 · Sample questions
10 real questions from the CertifyDeck DEA-C01 bank, spread across the domains of the official AWS Certified Data Engineer - Associate skills outline, with the correct answer and a full explanation of why the other options are wrong. They are original questions written from the public syllabus, not exam dumps, and they are the same style you will find in the practice tests and mock exams.
Correct answer: D. Migrate to Amazon RDS for PostgreSQL and enable IAM database authentication for the application.
On Amazon EC2 (infrastructure as a service) the customer patches the guest operating system, whereas Amazon RDS applies operating system updates as part of its maintenance. RDS for PostgreSQL also supports IAM database authentication, where the application uses an authentication token and its instance profile credentials instead of a password. Keeping passwords in Secrets Manager moves the database to a managed service but still relies on a password, which the requirement rules out.
Correct answer: D. Attach a dynamic data masking policy that partially redacts the card number column for the support role.
Redshift dynamic data masking changes how sensitive data is shown at query time without transforming it in the database, and masking policies can partially redact a column for specific roles while other roles see the full value, with no change to SQL queries. Excluding the column with a column-level grant hides it completely, so support users could not see the last four digits. A masked copy of the table adds a second dataset and forces the support queries to target a different table.
Correct answer: A. Run the procedure inside a transaction block that begins with a LOCK command on the table.
In Amazon Redshift, LOCK takes an ACCESS EXCLUSIVE table lock that makes reads and writes from other sessions wait, and it is released when the transaction that contains it completes, so it must run inside a transaction block. SNAPSHOT isolation still lets other sessions read a committed snapshot of the table, so it does not block access. Terminating or cancelling sessions only removes current users and does not stop new sessions from reading or loading the table.
Correct answer: D. analytics/sales/%
For engines that support schemas, such as Amazon Redshift, the JDBC include path is database/schema/table, and the percent sign can replace the schema or the table, so analytics/sales/% selects every table in that schema only. With Amazon Redshift, analytics/% creates tables for all schemas in the database. The percent sign cannot replace the database name, and sales/% would treat sales as the database.
Correct answer: C. Partition by day only, store the data in Parquet, and keep the records sorted by timestamp in each file.
Athena guidance is to partition for the most common queries rather than rare ones: partitioning by day and keeping records sorted by timestamp in a columnar format is almost as efficient for hourly filters and avoids the many small files that too many partition keys create. A minute-level key fragments the data further, and a partition index speeds up partition lookups but does nothing about the small files that hurt the common day-range queries.
Correct answer: A. Configure an incremental refresh on the order date column with a look-back window of a few days.
For SQL-based sources such as Amazon Redshift, an incremental refresh queries only the rows inside a look-back window defined on a date column, deletes the SPICE data in that window, and replaces it with the inserts, deletions and changes from the source. A full refresh reloads every row no matter how often it runs. Manifest files apply to Amazon S3 datasets, and direct query mode removes the data from SPICE, which breaks a stated requirement.
Correct answer: D. Change the sampling type of the project to a random selection of rows.
A DataBrew project works on a sample of the dataset, by default the first rows, so a file ordered by date shows only the oldest records. A random sample draws rows from across the dataset and better represents it. A larger first-n sample still starts at the oldest rows, a recipe job does not change the project sample, and limiting the dataset to the latest month makes the sample unrepresentative in the other direction.
Correct answer: A. Read UnprocessedItems from each response and retry those items with exponential backoff between the retry attempts.
A batch operation tolerates the failure of individual requests and returns them in UnprocessedItems, most often because of throttling, so the job must retry those items, and DynamoDB strongly recommends exponential backoff because immediate retries can be throttled again. Resending the whole batch at once repeats successful writes and keeps hitting the same capacity limit. TransactWriteItems makes a group all-or-nothing but does not remove throttling, and disabling retries makes failures more likely.
Correct answer: C. Set the execution class of these jobs to flexible so that they run as Flex job runs.
The flexible execution class is intended for non-urgent jobs such as pre-production jobs, testing and one-time data loads, and it is supported on AWS Glue 3.0 or later with G.1X or G.2X workers, so it fits these jobs and lowers their cost. The standard execution class is meant for time-sensitive workloads that need fast startup and dedicated resources, so it does not reduce cost, and a shorter timeout only risks failed runs. Larger G.8X workers and continuously running streaming jobs add capacity and cost instead of reducing it.
Correct answer: D. The Task uses Request Response rather than the Run a Job (.sync) pattern.
With the Request Response pattern, Step Functions moves to the next state as soon as it receives the HTTP response to StartJobRun, so validation starts while the job is still running. Using arn:aws:states:::glue:startJobRun.sync applies the Run a Job pattern, which waits for the job to complete. Standard workflows do support .sync for AWS Glue; only Express workflows are limited to Request Response. Job bookmarks only track processed data, and a missing permission would make the task fail rather than advance early.
Answer each question before opening the explanation, and write down why you discarded the other options. On DEA-C01, most mistakes come from choosing an option that is technically possible but does not meet every requirement in the scenario.
These 10 questions are a taste of the format, not a diagnostic. To know which domains you fail, practice by domain for free and take a timed mock exam that mirrors the official weights and duration.
Start for free, no sign-up or card. Every question includes an explanation like the ones above, and the mock exam reproduces the official DEA-C01 format.