SnowPro Advanced: Architect free question 8 / 10 · Domain 3: Data Engineering
A transient table DEV_SALES is created for development with CREATE TRANSIENT TABLE ... CLONE from the permanent table SALES_HIST (about 2 TB), and only some rows of DEV_SALES have been rewritten so far. SALES_HIST is later dropped with DROP TABLE, and its Time Travel retention period passes. Which description of how the micro-partitions that were shared with DEV_SALES are handled is consistent with the official documentation?
- They go into Fail-safe as soon as the retention period of SALES_HIST ends, and DEV_SALES can no longer read the rows it shared
- The shared bytes do not enter Fail-safe until DEV_SALES is dropped, and until then they remain counted as storage of DEV_SALES
- The shared bytes go into Fail-safe when the retention period of SALES_HIST ends, but they remain readable as rows of DEV_SALES
- Because DEV_SALES is transient and has no Fail-safe, the shared bytes are removed without ever going through Fail-safe
Show the answer and explanation
Answer: B
Why this is correct
Cloning a permanent table into a transient table creates a zero-copy clone that shares all the micro-partitions of the source table at the time it is created. When rows are added, deleted, or modified in the clone, micro-partitions that belong to the clone alone are created. The documentation says that dropping a permanent table sends its data into the seven-day Fail-safe, but when a transient clone shares micro-partitions with the permanent table, the shared bytes enter Fail-safe only when the transient table is dropped. Until DEV_SALES is dropped, the shared bytes therefore remain as storage of DEV_SALES, and the time they enter Fail-safe is delayed.
Why the other options are wrong
"They go into Fail-safe as soon as the retention period of SALES_HIST ends, and DEV_SALES can no longer read the rows it shared" … The documentation says the bytes shared with a transient clone enter Fail-safe when the transient table is dropped. Dropping the source table does not send them to Fail-safe, and the clone can still read them.
"Because DEV_SALES is transient and has no Fail-safe, the shared bytes are removed without ever going through Fail-safe" … The documentation says the shared bytes also enter Fail-safe when the transient table is dropped. Because the source is a permanent table, the bytes do not skip Fail-safe; they enter it later, once the transient table is dropped.
"The shared bytes go into Fail-safe when the retention period of SALES_HIST ends, but they remain readable as rows of DEV_SALES" … The shared bytes enter Fail-safe when the transient table is dropped, not when the retention period of the source table ends.
Key point
Bytes in Fail-safe are charged for storage as well. When you make a transient clone of a permanent table for development, Fail-safe for those bytes does not start until you drop the transient clone, even if you drop the source table. Include the period during which you keep the clone in your cost estimate.