Study SnowPro DEA-C02 Snowpipe Ingestion: key concepts, common traps, and exam decision cues.
DEA-C02 likes to compare staged-file automation with lower-latency ingest and then layer schema drift or semi-structured parsing on top. The winning answer usually starts with the latency target and source shape.
| Requirement | Better first instinct |
|---|---|
| automate staged-file arrival into Snowflake | Snowpipe |
| lower-latency streaming writes into Snowflake | Snowpipe Streaming |
| JSON or semi-structured payloads that need flexible parsing | semi-structured load strategy with VARIANT or controlled extraction |
| source columns reorder or grow over time | schema-handling or evolution decision at the load boundary |
| If the pressure is… | Stronger first answer |
|---|---|
| unattended staged-file arrival | Snowpipe |
| lower-latency ingest path | Snowpipe Streaming |
| flexible payload structure | semi-structured loading strategy |
| compatible column drift at load time | schema-handling decision |
| If the stem says… | Strong reading |
|---|---|
| “files arrive continuously in cloud storage” | Snowpipe is probably the first ingest lane |
| “low-latency ingestion” | Snowpipe Streaming becomes more relevant |
| “JSON already stored in VARIANT” | the next issue may be query logic, not reloading |
| “columns reorder or new nullable columns appear” | loading logic needs to tolerate compatible drift |
Both are ingestion answers, but they solve different operational shapes:
If the latency target is ordinary, the lower-latency answer can be unnecessary complexity.
| Trap | Better rule |
|---|---|
| choosing Snowpipe Streaming when staged-file automation is enough | match the ingest path to the latency need |
| blaming ingestion when the semi-structured query path is wrong | if the data is already loaded, check traversal logic first |
| tolerating schema drift by dropping rows silently | the load strategy should preserve correctness and visibility |
| Scenario clue | Stronger answer shape |
|---|---|
| “files arrive continuously in cloud storage” | Snowpipe |
| “lower-latency streaming writes into Snowflake” | Snowpipe Streaming |
“JSON payloads are already stored in VARIANT” |
query logic may matter more than reload logic |
| “columns reorder or new nullable columns appear” | load-boundary schema handling |
This lesson usually tests whether the ingest requirement is automated file loading or lower-latency streaming writes. If files arrive continuously in cloud storage, Snowpipe is usually the first answer. If the latency target is tighter and the source behavior is more stream-like, Snowpipe Streaming becomes stronger. The weak answer usually chooses the more advanced ingest path by default instead of matching it to latency and source behavior.