DEA-C01 Data Operations and Support Practice Question

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Configure the job to write output to a single file to reduce the number of S3 PUT requests.

  Why it's wrong here

  Writing to a single file reduces the number of output files but can create a bottleneck and limit parallelism during the write phase. It may also cause memory issues if the data is large. This approach contradicts the goal of improving performance through parallelism. It is better to write multiple partitioned files to leverage concurrent writes.

• ✓

  Convert the input data to a partitioned format, such as Parquet, and store it in S3 with a partition key.

  Why this is correct

  Partitioning the data and using a columnar format like Parquet allows Glue to read only relevant partitions and leverage predicate pushdown. This reduces I/O and enables parallel processing across partitions. It directly addresses the slow performance caused by a single large file by splitting data into manageable chunks that can be processed concurrently.

• ✗

  Enable job bookmarks to track processed data and avoid reprocessing.

  Why it's wrong here

  Job bookmarks help with incremental processing by tracking previously processed data, but they do not directly improve performance for a single large file. They reduce redundant work in subsequent runs but do not address parallelism or partitioning for the current dataset. The scenario focuses on speeding up the current job, not incremental processing.

• ✓

  Increase the number of DPUs allocated to the Glue job to allow more parallel tasks.

  Why this is correct

  Increasing DPUs provides more compute resources, which allows Glue to run more concurrent tasks and process data in parallel. This is effective when the job is resource-bound. However, it should be combined with data partitioning to fully utilize the additional capacity. It directly improves performance by scaling the job's execution environment.

• ✗

  Use the Glue DynamicFrame instead of a DataFrame to enable automatic schema inference.

  Why it's wrong here

  DynamicFrame provides schema flexibility and handles inconsistencies, but it does not inherently improve performance for large datasets. In fact, DynamicFrames can have overhead compared to DataFrames. The scenario requires performance optimization through partitioning and parallelism, not schema handling. This choice does not address the root cause of slow processing.

Go deeper

Related to this question