Zero-copy cloning in Snowflake creates a metadata-only snapshot of databases, schemas, or tables, enabling rapid environment refreshes without duplicating data. However, not all dependent objects behave identically during cloning.
Pipes that reference external stages are not cloned because external stages depend on credentials and integrations that may not exist or be valid in the cloned environment (Answer C). This is a documented Snowflake behavior and an important consideration when cloning production environments into development.
Regarding privileges, a cloned database inherits privileges granted on child objects (such as schemas and tables), but it does not inherit privileges granted directly on the source database itself (Answer E). This design prevents unintended access escalation and ensures that security must be explicitly re-granted at the database level.
Understanding these nuances is essential for SnowPro Architect candidates when designing environment promotion strategies and managing data lifecycle processes using cloning.
=========
QUESTION NO: 40 [Performance Optimization and Monitoring]
Two SQL queries perform the same task, but one consistently outperforms the other in reports, even though the SQL logic appears equivalent.
What step ensures the comparison is accurate?
A. Eliminate compilation time metrics.
B. Ensure both queries use the same clustering keys.
C. Disable result cache usage for the faster query.
D. Suspend and resume the virtual warehouse between queries.
Answer: C
Snowflake’s result cache can cause misleading performance comparisons. If one query benefits from cached results while another does not, execution time metrics will not reflect true performance differences. Disabling result cache usage for the faster query ensures both queries execute fully and fairly (Answer C).
Suspending and resuming a warehouse clears the local data cache but does not necessarily control result cache usage. Compilation time is generally minimal compared to execution time and does not explain consistent performance differences. Clustering keys are unrelated if both queries access the same data.
This question reinforces a core SnowPro Architect principle: accurate performance testing requires controlling for caching effects. Architects must understand Snowflake’s caching layers to correctly interpret query performance metrics.
