The query profile is a source of invaluable information, but often hard to digest due to the amount of data available. In this release, we’ve re-designed it to make it easier to identify opportunities for optimization:

• Added visual indicators to easily distinguish between operator types
• Re-designed the operator nodes to surface relevant information. For example, you can quickly spot full table scans by looking at the partitions scanned bar added to the TableScan nodes.
• Summarized the key bottlenecks in the query in the "Operator Types" side panel. Use this to understand if your query is spending most of the time on joins, table scans, sorts, etc.

In addition to improving the overall UI, we’ve also highlighted the three key metrics when analyzing a query profile:

• Duration
• Data/partitions scanned
• Data spilled

Using the Node Metric dropdown, these metrics can be highlighted on both the nodes and in the Operator Types and Operations panels on the right to aid investigation. For example, to identify the operations incurring the most spillage, switch the node metric to ‘Spillage’, and use the Operations panel to navigate:

🚀 What else we’ve shipped

• 🪄 Added new performance metrics related to records processed to workloads Cost & Performance tab
• 🪄 Improved rendering of large error messages in run history
• 🐛 Resolved an issue that was preventing the rendering of dbt workloads with lots of steps.