Explanation:
Draining the old pipeline ensures all existing data is processed before switching to the new pipeline, preventing data loss, inconsistencies, or unacceptable latency increases.
Ultimate access to all questions.
You've developed an updated version of a Dataflow streaming pipeline that ingests data from Pub/Sub into BigQuery. The current production pipeline uses a 5-minute window for processing. How can you deploy the new version without risking data loss, inconsistencies, or exceeding a 10-minute increase in processing latency?
A
Cancel the old pipeline, then immediately start the new pipeline.
B
Update the existing pipeline with the new code directly.
C
Drain the old pipeline to process all existing data before starting the new one.
D
Take a snapshot of the old pipeline, stop it, and start the new pipeline from the snapshot.
No comments yet.