Amazon Redshift announces enhancements to Advisor sort and distribution key recommendations

Amazon Redshift announces enhancements to Redshift Advisor to generate smarter and faster sort and distribution key recommendations. With the enhancements, Redshift Advisor uses new machine learning models to make these recommendations sooner, without needing to observe minimum necessary workload. Both sort and distribution key recommendations continue to be applied through Redshift automatic table optimizations, making it easier and faster to achieve better query performance.

Choosing the right sort key accelerates performance of queries by enabling the data warehouse to look through as little data as possible to serve the query. Choosing the right distribution key minimizes data movement across the nodes and helps manage data skew. Manually tuning sort and distribution keys effectively can be challenging, especially for large and complex workloads. To help with this, Redshift Advisor already provides intelligent sort and distribution key recommendations by analyzing your workloads and primary keys. The recommendations can sometimes take from a few hours up to days to be generated until enough workload is observed. Now, with these enhancements, Redshift Advisor recommends optimal sort and distribution keys sooner by leveraging new machine learning models to analyze column features such as column names, data types, and statistics, without needing to observe minimum necessary workload. Redshift Advisor will continue to monitor any workload changes and update sort and distribution key recommendations.

Enhancements to Amazon Redshift Advisor are available for Redshift provisioned and serverless data warehouses in all regions where Redshift Advisor is available. For more information about sort and distribution keys see our documentation.