2022 Performance Improvements


Performance improvements often target specific query patterns or workloads. These improvements might or might not have a material impact on a specific workload.

The following performance improvements were introduced in 2022.




November 2022

Improved performance and resilience of Snowflake applications and content.

Improved response times and availability for Snowsight pages.

November 2022

Ability to use scalar Vectorized Python UDFs in Snowpark.

Performance improvements for Python code that operates efficiently on batches of rows. Also requires less transformation logic when using Pandas DataFrames and arrays.

November 2022

Improvements for metadata queries.

Improved query execution time for small/metadata queries.

October 2022

Ability to enable Search Optimization for specific columns. (Preview)

Point lookup queries that act upon a column can be improved without incurring the expense of enabling Search Optimization for the entire table.

October 2022

Support for substring operations when using Search Optimization. (Preview)

Improves the performance of point lookup queries that use substring operations such as LIKE and ENDSWITH.

October 2022

Support for VARIANT data when using Search Optimization. (Preview)

Improves the performance of point lookup queries that act upon VARIANT data (such as JSON).

October 2022

Support for geospatial functions with GEOGRAPHY objects when using Search Optimization. (Preview)

Improves the performance of point lookup queries that use a geospatial function in a predicate.

October 2022

Improvements for Collation and BINARY columns.

Improved pruning for collations and BINARY columns, which means fewer micro-partitions must be scanned to return results.

October 2022

Improvements for hash table joins.

Improved query performance by reducing memory I/O latency in hash table equality checks.

October 2022

Improvements for DateTrunc range derivations.

Improved execution time for queries that use DateTrunc range derivation when TIMESTAMP-TZ data is a constant.

August 2022

Improvements related to Data Governance features.

More responsive Data Governance UI pages in Snowsight as well as improved query latency for tag-based masking policies.

August 2022

Improvements for window functions.

Improved rule-based optimization as well as improved query execution for outer join and filter pushdown in window functions.

August 2022

Scheduling improvements for high-concurrency workloads.

Improved query scheduling for high concurrency and lower latency workloads.

July 2022

Improved query performance using Join Elimination.

Optimized query performance through the automatic elimination of unnecessary joins, which are identified by automatically evaluating query logic.