A Developer is having a performance issue with a Snowflake query. The query receives up to 10 different values for one parameter and then performs an aggregation over the majority of a fact table. It then
joins against a smaller dimension table. This parameter value is selected by the different query users when they execute it during business hours. Both the fact and dimension tables are loaded with new data in an overnight import process.
On a Small or Medium-sized virtual warehouse, the query performs slowly. Performance is acceptable on a size Large or bigger warehouse. However, there is no budget to increase costs. The Developer
needs a recommendation that does not increase compute costs to run this query.
What should the Architect recommend?
Enabling the search optimization service on the table can improve the performance of queries that have selective filtering criteria, which seems to be the case here. This service optimizes the execution of queries by creating a persistent data structure called a search access path, which allows some micro-partitions to be skipped during the scanning process. This can significantly speed up query performance without increasing compute costs1.
Reference
* Snowflake Documentation on Search Optimization Service1.
Currently there are no comments in this discussion, be the first to comment!