Question

A data engineering team publishes an Amazon QuickSight dashboard built on a SPICE dataset that ingests from an Athena query over an S3 data lake. A nightly Glue ETL job finishes loading new partitions at 03:00. Business users complain that the dashboard shows the previous day's numbers every morning. The team confirms the Glue job completed successfully and Athena returns current data when queried directly. What is the MOST appropriate action to ensure the dashboard reflects the latest data each morning?

Accepted Answer

Schedule a SPICE dataset refresh to run after the Glue ETL job completes (for example at 03:30) so the imported data is updated before users log in.

Answer

Switch the QuickSight dataset from SPICE to direct query mode, which is the only way to guarantee freshness.

Answer

Recreate the Athena table with partition projection so QuickSight automatically detects new partitions.

Answer

Increase the SPICE capacity allocated to the account so the dataset can hold more rows.