How is query load calculated for an interval in Snowflake?

• A. Sum of execution times of all queries
• B. Total execution time of all queries divided by interval length
• C. Number of queries divided by interval length
• D. Average query execution time

Answer: (B)

The key idea is that query load is a time-based utilization measure. It reflects how much work the system handles within a given window by summing how long all queries ran and then normalizing that by the length of the window.

So, you take the total execution time of every query that executed during the interval and divide that sum by how long the interval lasts. That yields a rate that captures both how many queries ran and how long they occupied resources. For example, if two queries each ran for 60 seconds within a 60-second window, the total execution time is 120 seconds, and dividing by 60 seconds gives a load of 2. This indicates parallel activity: on average, there were two seconds of work happening per second of real time.

This approach distinguishes load from simply counting queries or just looking at average query duration. It combines duration and concurrency into a single, time-normalized metric.