Explanation:
The Query Profile in Snowflake provides an 'Operator Nodes by Execution Time' panel that lists nodes in descending order of execution time, specifically showing nodes that consumed 1% or more of the total query execution time. This allows users to quickly identify the most expensive operators. Option B directly references this functionality by suggesting to find the operator node with the highest fraction of time or percentage of total time. The community discussion confirms this approach with 100% consensus on answer B, citing Snowflake documentation that supports this method. Other options are less suitable: A focuses on micro-partitions scanned rather than execution time, C is too specific to TableScan operators and cache metrics, and D examines row counts between nodes which doesn't directly indicate operator cost in terms of execution time.
How can the Query Profile be utilized to identify the most expensive operator in a query?
A
Select any node in the operator tree and look at the number of micro-partitions scanned.
B
Find the operator node with the highest fraction of time or percentage of total time.
C
Select the TableScan operator node and look at the percentage scanned from cache.
D
Look at the number of rows between operator nodes across the operator tree.
No comments yet.