DVA-C02 · Question #779
A company runs a processing job every night to calculate customer loyalty scores. The job processes customer records sequentially. The job loads a customer's data from an Amazon DynamoDB table and…
The correct answer is D. Run multiple processing jobs in parallel. Use internal code logic to determine which records. The primary bottleneck is the 20-second ML computation per customer, which makes sequential processing slow and increasingly unable to finish overnight. Running multiple jobs in parallel reduces total elapsed time by distributing customers across concurrent workers while…
Question
A company runs a processing job every night to calculate customer loyalty scores. The job processes customer records sequentially. The job loads a customer's data from an Amazon DynamoDB table and runs a machine learning (ML) script to calculate a loyalty score for the customer. The job writes the output to a second DynamoDB table. The ML script takes 20 seconds to run. A developer observes an increase in the job duration. Some output results are not available the morning after the job runs. The developer does not see errors in the logs for DynamoDB. The developer needs to speed up the processing job. Which solution will meet this requirement with the LEAST amount of development effort?
Options
- AProvision read and write capacity units for both DynamoDB tables.
- BMigrate the ML script to Amazon SageMaker AI. Ref actor the processing job to query the
- CRefactor the processing job to read and write records by using the BatchGetItem and
- DRun multiple processing jobs in parallel. Use internal code logic to determine which records
How the community answered
(49 responses)- A4% (2)
- B22% (11)
- C14% (7)
- D59% (29)
Explanation
The primary bottleneck is the 20-second ML computation per customer, which makes sequential processing slow and increasingly unable to finish overnight. Running multiple jobs in parallel reduces total elapsed time by distributing customers across concurrent workers while leaving the existing DynamoDB access pattern largely unchanged. This achieves the speedup with minimal refactoring compared to redesigning the data access pattern or rebuilding the ML execution path.
Community Discussion
No community discussion yet for this question.