Skip links

Improving performance with Snowflake Clustering

Here is an example of performance improvements we’ve obtained using Snowflake Clustering.

Snowflake clustering

The before join between a 9.4 TB, 35.4 billion row table with a 1.5 TB, 1.8 billion row table ran in 1:14:57 and costs $11.99 in compute credits.

After clustering both tables, the tables automatically shrank in size due to improved columnar compression saving in storage costs and ran in 0:01:12 at a cost of $3.07, a savings of $8.92.

There is the added cost of clustering tables, which may or may not be cost effective for a given scenario. In addition to Clustering the Search Optimization Service and Query Acceleration Service are examples of Snowflake Performance Tuning possibilities.

Snowflake documentation