GCP-ADP Practice Test 15 — 15 Questions

Question 1

In Looker, what is the impact of defining a 'primary_key' in a LookML view?

Accepted Answer

It helps Looker avoid fan-outs during joins.. Defining a primary_key in a LookML view tells Looker that the field uniquely identifies each row in that view. During joins, Looker uses this information to determine the join relationship and can avoid fan-out (row duplication) by ensuring that the join is a many-to-one or one-to-one relationship. This prevents incorrect aggregation and improves query performance by allowing Looker to use symmetric aggregates.

Answer

It forces the database to create an index.

Answer

It changes the data type to an integer.

Answer

It restricts users from filtering by that column.

Question 2

Which TWO of the following are true about Workflows' serverless nature?

Accepted Answer

Scales automatically. Option B is correct because Workflows is a fully managed serverless orchestration service that automatically scales its execution capacity with demand, so you never provision or size workers yourself. Option C is correct because, as a serverless offering, Google manages all underlying infrastructure — servers, scaling, and availability — leaving you with no infrastructure to manage. Option A is wrong because Workflows does not require a dedicated GKE cluster; it runs on Google-managed infrastructure and can call GKE-hosted services but does not depend on a cluster. Option D is wrong because Workflows is not always free — it is billed per step executed, with a free tier only up to certain monthly limits. Option E is wrong because patching is handled by Google, not manually by the user, which is a core benefit of the serverless model.

Answer

Requires a dedicated GKE cluster

Answer

Is always free

Answer

Requires constant manual patching

Question 3

A team wants to display a BigQuery query result inside a Google Sheets spreadsheet so non-technical staff can view the latest figures. The data must refresh automatically each morning without anyone opening the sheet and clicking a refresh button. What should the team configure?

Accepted Answer

Use the Connected Sheets feature with a scheduled refresh of the BigQuery data source.. Connected Sheets maintains a live connection from a Sheets tab to a BigQuery data source and supports scheduled refreshes, so the figures update each morning automatically. Static pastes, CSV exports, and emailed PDFs all fail to deliver unattended, continuously current data in the spreadsheet.

Answer

Create a BigQuery scheduled query that emails the results as a PDF.

Answer

Export the BigQuery table to a CSV file in Cloud Storage and link the sheet to that file.

Answer

Paste the query results as static values into the sheet each morning.

Question 4

Which THREE of the following are key benefits of using Cloud Composer for data orchestration?

Accepted Answer

Uses standard open-source Airflow. Option A is correct because Cloud Composer is built on the open-source Apache Airflow project, so users can author DAGs with standard Airflow operators, hooks, and Python APIs without vendor-specific rewrites. Option D is correct because Cloud Composer is a fully managed service: Google provisions and maintains the Airflow environment, including the scheduler, web server, workers, and database, plus upgrades and patching. Option E is correct because Cloud Composer provides native integrations with GCP services such as BigQuery, Cloud Storage, Dataflow, Dataproc, and Pub/Sub through built-in operators and connections. Option B is not a key benefit of Composer itself, since transformation is performed by the tasks/services the DAGs invoke rather than by the orchestration layer. Option C is incorrect because Cloud Composer is not free for all users; it is a paid service billed for the environment's underlying compute, storage, and networking resources.

Answer

Automatic data transformation

Answer

Free tier for all users

Question 5

Which TWO of the following statements about Dataplex discovery are true? (Choose two)

Accepted Answer

Discovery scans and infers schema from the data. Option D is correct because Dataplex discovery automatically scans data in supported storage systems (such as Cloud Storage buckets and BigQuery datasets) and infers the schema by sampling and profiling the underlying files or tables. Option E is correct because the discovery process produces and publishes metadata — including schema, data profiles, and other statistics — into the Dataplex Data Catalog for the discovered assets. Option A is incorrect because discovery can run on a schedule (and can also be triggered on-demand), so it does not require manual triggering for every scan. Option B is incorrect because discovery supports multiple data sources, including Cloud Storage objects and BigQuery, not only BigQuery tables. Option C is incorrect because discovery is a metadata and schema inference service; it does not delete or deduplicate files.

Answer

Discovery must be manually triggered for every scan

Answer

Discovery works only for BigQuery tables

Answer

Discovery deletes duplicate files

Question 6

Which TWO of the following are valid ways to visualize BigQuery data within the Google Cloud ecosystem?

Accepted Answer

Looker Studio. Looker Studio (option C) is a valid visualization tool because it natively connects to BigQuery as a data source, letting you build interactive dashboards and reports directly from BigQuery tables or custom SQL queries. Looker (option E) is also correct since it is Google Cloud's enterprise BI platform that uses LookML models to query BigQuery and render dashboards, charts, and embedded analytics. The other options are not visualization tools: Cloud Storage (A) is an object storage service, Pub/Sub (B) is an asynchronous messaging service, and Bigtable (D) is a NoSQL wide-column database, none of which provide data visualization capabilities.

Answer

Cloud Storage

Answer

Pub/Sub

Answer

Bigtable

Question 7

Which THREE of the following are common administrative tasks when managing a Dataplex Lake? (Choose three)

Accepted Answer

Adding and removing data assets. Option A (Adding and removing data assets) is correct because a Dataplex Lake organizes zones and assets, and administrators routinely attach or detach assets (such as Cloud Storage buckets or BigQuery datasets) to make data discoverable and governed. Option D (Configuring data quality task schedules) is correct because Dataplex supports Data Quality tasks that can be scheduled to run on a recurring basis, and managing those schedules is a standard administrative duty. Option E (Managing IAM permissions for lake users) is correct because Dataplex uses Cloud IAM roles and policies (for example, roles/dataplex.admin, roles/dataplex.viewer) to control who can view, edit, or manage lakes, zones, and assets. Option B (Manually editing individual data files) is not a Dataplex administrative task, since Dataplex is a metadata and governance layer and does not edit underlying file contents. Option C (Rebuilding the physical database indexes) is not applicable, because Dataplex is a fully managed service and does not expose physical index maintenance to administrators.

Answer

Manually editing individual data files

Answer

Rebuilding the physical database indexes

Question 8

Which TWO actions can help reduce the time it takes to build a Jupyter notebook in Vertex AI Workbench?

Accepted Answer

Scaling up the machine type for the instance. Option A is correct because scaling up the machine type for the Vertex AI Workbench instance allocates more vCPUs and RAM, which directly speeds up compute-intensive notebook build and execution tasks such as model training or data processing. Option D is correct because using custom container images with pre-installed packages eliminates the time spent installing dependencies at notebook startup, reducing overall build time. Option B is incorrect because the project name has no effect on notebook build performance. Option C is incorrect because disabling the persistent disk would remove storage needed for the notebook environment and data, not improve build speed. Option E is incorrect because a single-core machine provides less compute capacity and would slow down, not reduce, build time.

Answer

Change the project name

Answer

Disable the persistent disk

Answer

Using a single-core machine

Question 9

Which THREE of these are valid ways to monitor or analyze BigQuery performance?

Accepted Answer

Query Plan explanation. Option A (Query Plan explanation) is correct because BigQuery's query execution plan, viewable in the Cloud Console or via the EXPLAIN statement, shows the stages, steps, and timing of a query so you can identify performance bottlenecks such as shuffles or expensive joins. Option B (INFORMATION_SCHEMA.JOBS_BY_PROJECT) is correct because this view exposes job metadata including query text, total bytes processed, slot milliseconds, and elapsed time, enabling historical performance analysis and cost attribution. Option E (BigQuery Monitoring dashboard in Cloud Console) is correct because it surfaces Cloud Monitoring metrics such as query execution time, slot usage, bytes processed, and concurrent jobs, giving real-time visibility into workload performance. Option C (Editing the table schema) is not a monitoring or analysis method; schema changes affect data structure and can influence performance but do not themselves observe or analyze it. Option D (Changing the project billing currency) is purely a billing configuration action and has no bearing on query performance monitoring or analysis.

Answer

Editing the table schema

Answer

Changing the project billing currency

Question 10

Which TWO of the following are valid locations for a Dataplex Lake? (Choose two)

Accepted Answer

US (Multi-region). Options D and E are correct because a Dataplex Lake is a regional or multi-regional Google Cloud resource, so it must be created in a supported Google Cloud location such as the US multi-region (D) or a specific region like us-central1 (E). Dataplex lakes are scoped to a Google Cloud region or multi-region, which determines where the lake's metadata and associated assets are stored and managed. Option A is incorrect because an on-premises data center is not a Google Cloud location and cannot host a Dataplex Lake, even though on-premises data can be attached via Dataplex assets. Option B is incorrect because the root organization node is an IAM/resource hierarchy scope, not a geographic location for a lake. Option C is incorrect because a user's local laptop is not a Google Cloud region or multi-region and cannot serve as a Dataplex Lake location.

Answer

A private on-premises data center

Answer

The root organization node

Answer

The user's local laptop

Question 11

Which TWO are methods to secure GCS data?

Accepted Answer

IAM roles. IAM roles (B) are a correct method to secure GCS data because Cloud Storage uses Cloud IAM to grant fine-grained, least-privilege access to buckets and objects, controlling who can read, write, or administer data via roles such as roles/storage.objectViewer or roles/storage.admin. Cloud KMS encryption (E) is also correct because Customer-Managed Encryption Keys (CMEK) let you encrypt GCS objects with keys you own and manage in Cloud KMS, giving you control over key rotation, revocation, and audit of cryptographic access. Renaming files (A) does not secure data, since object names are not a security control and access is governed by IAM and ACLs, not filenames. Printing files (C) is unrelated to data security and could actually increase exposure by creating physical or additional copies. Public internet access (D) is the opposite of securing data, as making a bucket public grants anonymous read access and violates the principle of least privilege.

Answer

Renaming files

Answer

Printing files

Answer

Public internet access

Question 12

A team is migrating to Dataplex. Which TWO of the following are prerequisites for adding a BigQuery dataset as an asset to a Lake? (Choose two)

Accepted Answer

The user needs BigQuery permissions on the dataset. Option A is correct because registering a BigQuery dataset as a Dataplex asset requires the user to hold BigQuery permissions on that dataset (for example, roles such as BigQuery Data Viewer/Owner or bigquery.datasets.get and bigquery.tables.get) so Dataplex can read the dataset's metadata and contents. Option B is correct because the user must also hold Dataplex asset-management permissions (such as dataplex.assets.create/update/delete, typically via roles/dataplex.admin or roles/dataplex.editor) to attach the dataset as an asset within the Lake. Option C is wrong because the dataset does not need to be public; access is governed by IAM permissions, and public exposure is neither required nor recommended. Option D is wrong because a BigQuery dataset can reside in a different project than the Lake, provided cross-project IAM permissions and the appropriate resource references are configured. Option E is wrong because an empty dataset is still a valid asset; Dataplex does not require the dataset to contain at least one table.

Answer

The dataset must be public

Answer

The dataset must exist in the same project as the Lake

Answer

The dataset must contain at least one table

Question 13

Which TWO of the following are GCP ingestion tools?

Accepted Answer

Cloud Storage. Cloud Storage (A) is correct because it serves as a GCP ingestion tool by allowing data to be loaded into the cloud via gsutil, the JSON/XML APIs, or Storage Transfer Service, and it commonly acts as the landing zone for batch data pipelines feeding BigQuery or Dataproc. Pub/Sub (D) is correct because it is GCP's fully managed, globally scalable messaging service that ingests streaming data from publishers and delivers it to subscribers, making it the primary ingestion mechanism for real-time and event-driven pipelines. Cloud SQL (B) is not an ingestion tool; it is a managed relational database service (MySQL, PostgreSQL, SQL Server) used for storage and querying, not for bringing data into GCP. Cloud Compute (C) is not an ingestion tool; Compute Engine provides IaaS virtual machines, which may run ingestion software but are not themselves an ingestion service. Cloud DNS (E) is not an ingestion tool; it is a scalable, reliable managed DNS service for publishing and resolving domain names.

Answer

Cloud SQL

Answer

Cloud Compute

Answer

Cloud DNS

Question 14

Which THREE of the following are valid methods for troubleshooting failed tasks in Cloud Composer?

Accepted Answer

Checking the Airflow web UI task state. Option A is correct because the Airflow web UI shows each task instance's state (queued, running, success, failed, up_for_retry, etc.), letting you quickly identify which tasks failed and drill into their details. Option D is correct because the Composer scheduler logs record DAG parsing errors, scheduling decisions, and task-queueing problems, which are essential for diagnosing why tasks never ran or were stuck. Option E is correct because task instance logs capture the actual stdout/stderr from the task's execution, revealing the specific exception or failure cause. Option B is not appropriate because rebuilding the entire GKE cluster is a drastic, disruptive action that destroys the environment rather than a troubleshooting step. Option C is not appropriate because deleting the environment bucket removes DAGs, logs, and data, causing data loss and breaking the environment instead of diagnosing the failure.

Answer

Rebuilding the entire GKE cluster

Answer

Deleting the environment bucket

Question 15

You need to organize your data governance strategy. Which TWO of the following are valid ways to use Dataplex Zones to improve your data organization? (Choose two)

Accepted Answer

Assign different IAM policies to different zones. Option A is correct because Dataplex zones are logical groupings of data assets within a lake, and IAM policies can be attached at the zone level, allowing you to grant different access permissions to different zones (e.g., restricting raw data while opening curated data to analysts). Option D is correct because zones are designed to organize data by its lifecycle or processing state, so creating raw, curated, and consumption zones is a canonical Dataplex pattern for data governance. Option B is not valid because query results are stored in BigQuery datasets or temporary tables, not in Dataplex zones, which are metadata/organization constructs rather than storage destinations. Option C is not valid because zones group data assets by governance or lifecycle characteristics, not by individual users; per-user access is handled through IAM, not by creating a zone per user. Option E is not valid because zones do not affect storage performance or speed; they are organizational and governance constructs, and storage speed is determined by the underlying service (e.g., Cloud Storage or BigQuery).

Answer

Use zones to store query results

Answer

Create separate zones for each user

Answer

Use zones to increase data storage speed