ADA-C01 Question 256
Single answerConfigure tasksA data engineering team wants to run a daily ETL pipeline in Snowflake using tasks. The pipeline has three steps: (1) load raw files into a staging table, (2) transform and merge the staged data into a curated table, and (3) refresh a reporting aggregate. The team needs step 1 to start every day at 01:00 UTC, and steps 2 and 3 must run only after their immediate predecessor completes successfully. They also want to minimize administrative overhead and avoid assigning a warehouse just to keep the schedule running. Which configuration best meets these requirements?
- A
Create three independent scheduled tasks, each with the same CRON schedule at 01:00 UTC, and assign all of them to a user-managed warehouse.
- B
Create a root task scheduled with CRON at 01:00 UTC using a serverless task, then create step 2 as an AFTER task dependent on step 1 and step 3 as an AFTER task dependent on step 2.
- C
Create a single task with a CRON schedule at 01:00 UTC that calls all three SQL statements separated by semicolons, and configure AFTER relationships for failover.
- D
Create step 1 as a task triggered by a stream on the staging table, then schedule step 2 and step 3 independently at 01:00 UTC using serverless tasks.
Show answer and explanation
Correct answer: B
Explanation
The best solution is to use a task graph with one scheduled root task and dependent child tasks defined with AFTER. In Snowflake, task graphs are designed for orchestrating dependent workflows where downstream tasks should execute only after upstream tasks succeed. For this scenario, the root task should be scheduled with a CRON expression for 01:00 UTC, and the subsequent tasks should be chained through AFTER dependencies. To minimize administrative overhead, serverless tasks are the best fit because Snowflake manages the compute resources automatically, eliminating the need to assign and maintain a user-managed warehouse solely for task execution. This approach aligns with Snowflake best practices for orchestrating ETL pipelines with clear dependencies, predictable scheduling, and lower operational burden.
- A. Incorrect.
Incorrect. Independent scheduled tasks set for the same time do not guarantee execution order. Snowflake task graphs should be used when one task must wait for another to finish successfully. In addition, using a user-managed warehouse does not meet the stated goal of minimizing administrative overhead related to compute management.
- B. Correct.
Correct. A task graph is the appropriate design for ordered ETL steps. The root task can use a CRON schedule to start the workflow at 01:00 UTC, and child tasks can be linked with AFTER so each runs only after its predecessor succeeds. Using serverless tasks avoids the need to provision and manage a virtual warehouse for task execution, which aligns with the requirement to reduce operational overhead.
- C. Incorrect.
Incorrect. While a task can execute SQL, combining the entire pipeline into one task reduces operational control, observability, and modular failure handling. More importantly, AFTER relationships are used between separate tasks in a task graph, not as a failover mechanism within a single task definition. This option does not represent the correct way to model dependent task execution in Snowflake.
- D. Incorrect.
Incorrect. A triggered task based on a stream is useful for event-driven processing when data arrival should initiate execution. However, the requirement explicitly states that step 1 must start daily at 01:00 UTC. Scheduling downstream steps independently at the same time also fails to guarantee that they run only after the previous step completes successfully.