GCP Professional Data Engineer Practice Question

Your team operates a streaming Dataflow pipeline that processes sales events 24/7. The job is launched with the flag --worker_zone=us-central1-a. A recent zonal outage in us-central1-a stopped all workers for several hours, breaching the SLA. You must change the launch configuration so the pipeline continues running if any single zone in the region fails, without maintaining a duplicate job or significantly increasing cost. Which action satisfies the requirement?

Containerize the pipeline and deploy it to Cloud Run in two regions behind Cloud Load Balancing.
Migrate the code to run on an HA Dataproc cluster with masters in three zones and submit the streaming job there.
Launch the job with --worker_region=us-central1 and remove the --worker_zone flag so Dataflow distributes workers across healthy zones in the region.
Keep the existing job and schedule a second, identical pipeline in us-central1-b; switch Pub/Sub subscriptions to whichever job is healthy.

Report Issue

Answer Description

When you specify only a region (for example --worker_region=us-central1) and omit any --worker_zone flag, Dataflow automatically provisions and balances worker VMs across multiple zones inside that region. If one zone becomes unavailable, workers in the remaining zones continue processing, and Dataflow can spin up replacement workers in healthy zones. Running a second, parallel pipeline or migrating to a different service would add unnecessary cost or operational complexity. Converting to a Dataproc or Cloud Run solution changes the service entirely rather than addressing the zonal-outage issue for the existing pipeline.

Ask Bash

Bash is our AI bot, trained to help you pass your exam. AI Generated Content may display inaccurate information, always double-check anything important.