Answer-first summary for fast verification
Answer: Drain the old pipeline to process all existing data before starting the new one.
Draining the old pipeline ensures all existing data is processed before switching to the new pipeline, preventing data loss, inconsistencies, or unacceptable latency increases. - **Option A** risks data loss as existing data may not be fully processed. - **Option B** could introduce inconsistencies during the transition due to differing processing logic. - **Option D** might still result in data loss or inconsistencies if the transition isn't meticulously managed.
Author: LeetQuiz Editorial Team
Ultimate access to all questions.
No comments yet.
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.