Window functions in Snowflake - Step-by-Step Execution

Process Flow - Window functions in Snowflake

Start Query

↓

Select Data

↓

Define Window

↓

Apply Window Function

↓

Calculate Result per Row

↓

Return Result Set

The query starts by selecting data, then defines a window (a group of rows), applies the window function to each row within that window, calculates results, and returns the final result set.

Execution Sample

Snowflake

SELECT
  employee_id,
  department,
  salary,
  AVG(salary) OVER (PARTITION BY department ORDER BY employee_id ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS avg_salary
FROM employees;

This query calculates the running average salary per department ordered by employee_id.

Process Table

Step	Row (employee_id)	Partition (department)	Window Frame	Window Function Calculation	avg_salary
1	101	Sales	Rows from start to current (101)	AVG(salary) of [5000]	5000
2	102	Sales	Rows from start to current (101,102)	AVG(salary) of [5000, 6000]	5500
3	103	Sales	Rows from start to current (101,102,103)	AVG(salary) of [5000, 6000, 6500]	5833.33
4	201	HR	Rows from start to current (201)	AVG(salary) of [4500]	4500
5	202	HR	Rows from start to current (201,202)	AVG(salary) of [4500, 4700]	4600
6	301	IT	Rows from start to current (301)	AVG(salary) of [7000]	7000
7	302	IT	Rows from start to current (301,302)	AVG(salary) of [7000, 7200]	7100
8	303	IT	Rows from start to current (301,302,303)	AVG(salary) of [7000, 7200, 7100]	7100
9	END	-	-	-	All rows processed, query returns result set

💡 All rows processed, window function applied per partition and ordered rows, query completes.

Status Tracker

Variable	Start	After 1	After 2	After 3	After 4	After 5	After 6	After 7	After 8	Final
employee_id	-	101	102	103	201	202	301	302	303	-
department	-	Sales	Sales	Sales	HR	HR	IT	IT	IT	-
salary	-	5000	6000	6500	4500	4700	7000	7200	7100	-
avg_salary	-	5000	5500	5833.33	4500	4600	7000	7100	7100	-

Key Moments - 3 Insights

Why does the average salary change only within each department and not across all employees?

Because the window function uses PARTITION BY department, it calculates averages separately for each department partition as shown in execution_table rows 1-3 for Sales, 4-5 for HR, and 6-8 for IT.

What does the window frame 'ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW' mean?

It means the function calculates the average from the first row in the partition up to the current row, accumulating values as seen in the execution_table where the window frame grows with each row.

Why is the order by employee_id important in the window function?

Ordering by employee_id defines the sequence in which rows are considered for the running average, affecting the calculation step by step as shown in the execution_table where avg_salary changes with each ordered row.

Visual Quiz - 3 Questions

Look at the execution_table at step 5. What is the avg_salary value for employee_id 202 in HR?

A 4500

B 4700

C 4600

D 4550

💡 Hint

Check the avg_salary column in execution_table row 5 for employee_id 202.

At which step does the window function finish processing all rows in the IT department?

A Step 8

B Step 7

C Step 6

D Step 9

💡 Hint

Look at the partition column and see when IT department rows end in execution_table.

If we remove ORDER BY employee_id from the window clause, how would the avg_salary values change?

A They would be calculated over the entire partition without order, so running average would not accumulate.

B The avg_salary would be the overall average per department for all rows.

C The query would fail with an error.

D They would be the same as now.

💡 Hint

Consider how ORDER BY affects the window frame and calculation in execution_table steps.

Concept Snapshot

Window functions in Snowflake:
- Use OVER() clause with PARTITION BY and ORDER BY
- PARTITION BY groups rows like departments
- ORDER BY defines row sequence inside partitions
- Window frame defines rows considered per calculation
- Functions compute values per row over the window
- Useful for running totals, averages, ranks without grouping

Full Transcript

This visual execution traces a Snowflake window function calculating running average salary per department. The query selects employee data and applies AVG() over a window partitioned by department and ordered by employee_id. Each step shows how the window frame grows from the first row to the current row in the partition, updating the average salary. Variables track employee_id, department, salary, and the computed avg_salary. Key moments clarify partitioning, window frame meaning, and ordering importance. The quiz tests understanding of values at specific steps and effects of removing ORDER BY. The snapshot summarizes syntax and behavior for quick reference.