PDE - Ingesting and Processing Data (25% of the exam) - Section 2.4

Deploy and operationalise pipelines using Cloud Composer and Workflows for job automation, and implement CI/CD for data pipelines.

Deploy data pipelines using Cloud Composer DAGs and Apache Airflow operators or Google Cloud Workflows for serverless step orchestration, and configure CI/CD pipelines to test and promote pipeline code across environments reliably.

Cloud ComposerWorkflowsCI/CDApache Airflow

Practice question for this objective

Free sampleIngesting and Processing Datamedium

A data platform team is choosing between Cloud Composer and Workflows to orchestrate their data pipelines. They have a daily batch flow with thirty interdependent tasks across BigQuery, Dataflow, and Dataproc, plus a need for backfills, conditional branching on data quality results, and rich operator libraries maintained by an open-source community. Which statement best characterises the role boundary between the two services for this workload?

AWorkflows is the better fit because it provides native operators for BigQuery, Dataflow, and Dataproc and supports backfills with a DAG scheduler maintained by the open-source community.
BCloud Composer and Workflows are interchangeable for batch data pipelines because both expose DAGs, backfills, and Airflow operators behind a managed runtime.
CWorkflows is the better fit because long-running, stateful data pipelines with thirty tasks are its primary use case, while Cloud Composer is intended for short HTTP service orchestrations.
DCloud Composer is the better fit because it runs managed Apache Airflow, which provides the operator ecosystem, DAG backfills, and branching primitives such as BranchPythonOperator that suit this multi-step data pipeline. Correct

Recognise that Cloud Composer is managed Apache Airflow for data pipeline orchestration, while Workflows targets lighter service and event orchestration. Cloud Composer is a managed Apache Airflow service, so it inherits Airflow's scheduler, DAG model, backfills, branching operators, and the large community library of Google Cloud operators. Workflows is a separate product for orchestrating HTTP services and Google Cloud connectors via a steps-based definition; it does not expose Airflow concepts such as DAG runs or backfills, so a thirty-task interdependent batch pipeline maps more naturally to Composer.

Why A is wrong: Workflows does have connectors for Google Cloud services, but it is not built on Apache Airflow, it has no concept of DAG backfills, and its connector catalogue is not an open-source community library. The candidate may be confusing Workflows connectors with Airflow operators.

Why B is wrong: The services are not interchangeable. Only Cloud Composer runs Airflow and supports DAG backfills and Airflow operators. Workflows uses YAML or JSON step definitions and a different execution model, so this conflates the two products.

Why C is wrong: This inverts the positioning of the two services. Workflows is optimised for lightweight, event-driven service orchestration with HTTP and connector steps, while Cloud Composer is the data-pipeline orchestrator built on Airflow for richer multi-task DAGs.

Why D is correct: Cloud Composer is managed Apache Airflow, and Airflow's strengths are exactly the requirements stated: a large community-maintained operator catalogue for Google Cloud services, first-class backfills via the scheduler, and conditional branching through operators such as BranchPythonOperator.

See more PDE practice questions, answers explained.

Exam traps in Ingesting and Processing Data

Answers that look right on this material and are not. Each one is a distractor from a different question in the PDE bank for this domain.

Have engineers continue editing DAGs in the production GCS bucket but enable object versioning so any broken change can be rolled back to the previous object version after the scheduler reports a failure.
Why it is wrong: Object versioning lets the team roll back after the fact, which is tempting because it gives some history, but it does not prevent broken DAGs from reaching the scheduler and there is no review, no testing and no separation between environments.
Workflows runs a long-lived Airflow scheduler and worker fleet, so it bills per node-hour just like Cloud Composer environments and offers no cost advantage for short, infrequent runs.
Why it is wrong: Workflows does not run Airflow and does not bill per node-hour. Cloud Composer is the service with a long-lived environment that incurs cost while idle. The candidate may be assuming both services share Composer's pricing shape.
Use a single Cloud Build trigger on pull requests that runs tests, parses the DAGs with python -c imports, and on success copies the DAG files to all three environment buckets simultaneously.
Why it is wrong: Copying to all three environment buckets at once bypasses the staging gate and the manual production approval the team explicitly requires, so it violates the promotion model even though the tests and parse checks themselves are appropriate.

Examworthy is not affiliated with or endorsed by Google Cloud. Original, blueprint-aligned practice material only.