Query profiling and the query plan in Snowflake - Step-by-Step Execution

Process Flow - Query profiling and the query plan

Submit SQL Query

↓

Query Parser

↓

Query Optimizer

↓

Generate Query Plan

↓

Execute Query Plan

↓

Collect Profiling Data

↓

Show Query Profile & Plan

The query is parsed, optimized, and a plan is created. The plan runs, collecting data. Finally, the profile and plan are shown.

Execution Sample

Snowflake

EXPLAIN USING TEXT SELECT * FROM orders WHERE amount > 100;
-- Then run
SELECT * FROM orders WHERE amount > 100;

Shows the query plan text, then runs the query to collect profiling data.

Process Table

Step	Action	Details	Result
1	Submit Query	SELECT * FROM orders WHERE amount > 100	Query received by Snowflake
2	Parse Query	Check syntax and semantics	Query parsed successfully
3	Optimize Query	Find best plan for filtering amount > 100	Query plan created with filter pushdown
4	Generate Query Plan	Plan includes scan of orders table with filter	Plan ready for execution
5	Execute Query Plan	Scan orders, apply filter amount > 100	Rows matching condition retrieved
6	Collect Profiling Data	Measure time, bytes scanned, rows processed	Profiling data collected
7	Show Query Profile & Plan	Display execution steps and resource usage	User sees detailed query plan and profile

💡 Query execution completes and profiling data is available for analysis

Status Tracker

Variable	Start	After Parsing	After Optimization	After Execution	Final
Query Text	SELECT * FROM orders WHERE amount > 100	Parsed	Optimized	Executed	Completed
Query Plan	None	None	Scan orders with filter	Used to fetch rows	Available for profile
Rows Processed	0	0	0	N rows matching filter	N rows returned
Bytes Scanned	0	0	0	X bytes scanned	X bytes recorded
Execution Time	0	0	0	Measured during execution	Total time recorded

Key Moments - 3 Insights

Why does the query plan show a filter operation before scanning all rows?

Because the optimizer pushes the filter down to reduce data scanned, as shown in step 3 and 4 of the execution table.

What does the profiling data tell us after query execution?

It shows resource usage like bytes scanned and execution time, helping understand query efficiency, as collected in step 6.

Why is the EXPLAIN command useful before running the query?

It shows the planned steps without running the query, so you can see how Snowflake will execute it, as in step 1 and 7.

Visual Quiz - 3 Questions

Look at the execution table, at which step is the query plan created?

A Step 4

B Step 3

C Step 2

D Step 5

💡 Hint

Check the 'Generate Query Plan' action in the execution table rows.

According to the variable tracker, when are rows processed counted?

A After Parsing

B After Optimization

C After Execution

D At Start

💡 Hint

Look at the 'Rows Processed' variable values in the variable tracker.

If the filter condition changes, how does the query plan step change?

A The plan stays the same

B The filter operation in the plan updates

C Parsing step changes only

D Profiling data changes only

💡 Hint

Refer to the 'Optimize Query' and 'Generate Query Plan' steps in the execution table.

Concept Snapshot

Query profiling shows how a query runs and uses resources.
The query plan is the step-by-step plan Snowflake makes to run the query.
Use EXPLAIN to see the plan before running.
Profiling data includes time, rows, and bytes scanned.
Optimizers push filters early to reduce work.
Profiling helps find slow or costly queries.

Full Transcript

When you submit a query in Snowflake, it first gets parsed to check for errors. Then the optimizer creates a plan to run the query efficiently. This plan includes steps like scanning tables and applying filters. When the query runs, Snowflake collects profiling data such as how many rows were processed, how many bytes were scanned, and how long it took. You can use the EXPLAIN command to see the query plan before running the query. This helps you understand how Snowflake will execute your query and can help you improve performance by adjusting your SQL or data. Profiling data after execution shows the actual resource usage and helps identify bottlenecks.