SnowPro Associate: Platform Question 356
Single answer○ Search listingsA data provider publishes a Snowflake Marketplace listing that includes a large customer table. Consumer accounts frequently run selective queries such as filtering by CUSTOMER_ID and EMAIL, but query performance is inconsistent because the table is very large. The provider wants to improve consumer query performance for this listing without requiring consumers to redesign their SQL. Which action should the provider take?
- A
Add a search optimization service on the relevant columns in the shared table before publishing or updating the listing
- B
Create a warehouse in the consumer account and attach it to the listing so Marketplace queries automatically use more compute
- C
Convert the listing to an application package because search listings only work with Native Apps
- D
Require consumers to create clustering keys on the shared table after installing the listing
Show answer and explanation
Correct answer: A
Explanation
For Marketplace and other shared data scenarios, the data provider controls the underlying physical design of the shared objects. If consumers commonly run selective predicates on large shared tables, the provider can enable the Search Optimization Service on appropriate columns to improve performance. This is especially useful for point-lookups and highly selective queries, and it does not require consumers to change ownership of the data or redesign their SQL. Consumers use their own virtual warehouses to run the queries, but they cannot alter shared tables with features such as clustering keys. Snowflake documentation on the Search Optimization Service and data sharing/listings emphasizes that optimization of shared objects must be configured by the provider on the source data.
- A. Correct.
Correct. Search optimization can be enabled by the provider on supported tables and columns to accelerate highly selective point-lookups and similar predicates. For data shared through a listing, consumers query the provider's shared data and cannot change the physical design of the provider's underlying objects. Enabling search optimization is a provider-side action that can improve performance for consumer queries without requiring query rewrites.
- B. Incorrect.
Incorrect. Compute for queries against shared data is still run in the consumer account using the consumer's warehouse, but a provider cannot create and attach a warehouse inside the consumer account through a listing. Also, simply adding more compute is not what a search listing addresses; the scenario is about optimizing access paths for selective filters.
- C. Incorrect.
Incorrect. This confuses Marketplace listings with Native Apps. Search optimization for shared data is not dependent on converting the offering to an application package. A standard listing can expose data that benefits from search optimization; there is no requirement that 'search listings' are only for Native Apps.
- D. Incorrect.
Incorrect. Consumers cannot define clustering keys on shared tables they do not own. Clustering is also a different optimization technique focused on micro-partition organization, whereas the scenario specifically involves improving selective lookup performance for data shared through a listing without requiring consumer-side changes.