2023 Performance Improvements¶

Important

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 2023.

Released	Description	Impact
December 2024	Improved column replication.	Reduces the time spent in the SECONDARY_DOWNLOADING_METADATA phase of a refresh operation for table columns. Improvements scale linearly with the number of columns replicated.
November 2023	Improved execution times for some SHOW commands.	Reduces the execution time for the SHOW TABLES, SHOW SCHEMAS, and SHOW DATABASES commands. Improvements are most significant for queries that return large result sets.
November 2023	Search Optimization: Support for substring search in semi-structured data. (General Availability)	Improves the performance of point lookup queries that use substring and regular expression functions against semi-structured data, including ARRAY, OBJECT, and VARIANT types. Previously, only equality searches on such columns could be optimized.
October 2023	Reduced maintenance costs for materialized views.	Reduces materialized view maintenance credits by improving the utilization of service resources.
October 2023	Improved compile times for SQL expressions.	Reduces the compilation time of queries that contain many SQL expressions.
September 2023	Improved compile times.	Reduces compilation times by skipping optimizations that will not result in performance improvements.
August 2023	Ability to use a query hash to identify patterns and trends in query execution.	Helps to monitor and analyze recurring queries by including a query hash of each query in ACCOUNT_USAGE views and INFORMATION_SCHEMA table functions. Can be used to determine the effects of performance improvements like choosing a new cluster key.
August 2023	Improved execution times for non-clustered tables.	Reduces execution time for SELECT and DML operations against non-clustered tables with micro-partitions that are smaller than average.
August 2023	Ability to call the GET_QUERY_OPERATOR_STATS function to obtain query profile statistics. (General Availability)	Helps to programmatically debug queries and gain insights into query performance.
August 2023	Improved execution times for joins on wide build-side rows.	Reduces execution time and improves memory management for queries matching wide rows on the build side of a join (for example, rows that include columns with long strings).
July 2023	Improved compile times for materialized views.	Reduces compilation times for materialized views based on tables that have 100s or 1000s of micro-partitions.
July 2023	Ability to use Snowpipe Streaming. (General Availability)	Enables low-latency streaming data pipelines to support writing data rows directly into Snowflake.
July 2023	Improved selectivity and cardinality estimation.	Uses improved plan selection to reduce execution time for queries with low selectivity.
July 2023	Search Optimization Update: Support for substring search in VARIANT types. (Preview)	Improves the performance of point lookup queries that use substring and regular expression functions against semi-structured data, including ARRAY, OBJECT, and VARIANT types.
July 2023	Improved compile times for simple queries and DML statements.	Reduces compilation times and improves memory management for simple DML statements and single-table queries with simple equality or range predicates.
June 2023	Improved execution times for SELECT statements with LIMIT and ORDER BY clauses.	Reduces execution time of some queries with long-running SELECT statements containing both LIMIT and ORDER BY clauses.
June 2023	Improved execution times against secure views.	Uses predicate pushdown to reduce execution time for queries against secure views.
May 2023	Improved compile times for queries with numerous extraction expressions.	Reduces compilation times for queries with many extraction expressions (such as those used for processing JSON).
May 2023	Improved compile times for queries with numerous subqueries.	Reduces compilation time for queries with 100+ subqueries.
April 2023	Search Optimization Update: Ability to enable Search Optimization for specific columns. (General Availability)	Point lookup queries that act upon a column can be improved without incurring the expense of enabling Search Optimization for the entire table.
April 2023	Search Optimization Update: Support for substring operations. (General Availability)	Improves the performance of point lookup queries that use substring operations such as LIKE and ENDSWITH.
April 2023	Search Optimization Update: Support for VARIANT data. (General Availability)	Improves the performance of point lookup queries that act upon VARIANT data (such as JSON).
April 2023	Search Optimization Update: Support for geospatial functions with GEOGRAPHY objects. (General Availability)	Improves the performance of point lookup queries that use a geospatial function in a predicate.
April 2023	Ability to use the query acceleration service to speed up queries against tables that have Search Optimization enabled. (General Availability)	Additional compute power provided by the query acceleration service can be combined with performance boost provided by search optimization.
March 2023	Ability to use Snowpipe Streaming. (Preview)	Enables low-latency streaming data pipelines to support writing data rows directly into Snowflake.
February 2023	Ability to use the query acceleration service. (General Availability)	Improves overall warehouse performance by reducing the impact of outlier queries.
February 2023	Ability to call the GET_QUERY_OPERATOR_STATS function to obtain programmatic Query Profile statistics. (Preview)	Helps debug queries and gain insights into query performance.
February 2023	Ability to use memory-optimized warehouses.	Memory-intensive queries can be run on Snowpark-optimized warehouses that provide 16x more memory per node and 10x the local cache compared with standard warehouses.