Free Google Cloud Professional Data Engineer standard exam Practice Questions

Sample Practice Questions

Try these 10 sample questions from the Google Cloud Professional Data Engineer standard exam practice bank. Each includes the answer and explanation. Continue with the interactive practice above to review the full question set.

1. 1A BigQuery table contains sales from several regions. Each regional group must query only its own rows without maintaining separate copies. Which control should the engineer implement?

   1. A. Column policy tags on the region column
   2. B. Row access policies with regional filter expressions
   3. C. Dynamic masking on the sales amount column
   4. D. A service perimeter around the BigQuery project

   Answer: B

   Explanation: The documentation states, "Row-level security lets you filter data and enables access to specific rows in a table based on qualifying user conditions." Regional row filters meet the requirement; column controls and service perimeters do not select each group’s rows.

   Reference: Introduction to BigQuery row-level security — What is row-level security?

2. 2Dataflow worker VMs have only internal IP addresses. Their subnet must allow access to Google Cloud APIs without assigning external addresses. Which subnet feature directly supports this access?

   1. A. Private Google Access
   2. B. Cloud NAT
   3. C. VPC Network Peering
   4. D. A default route without Private Google Access

   Answer: A

   Explanation: The guide states, "With Private Google Access, VMs that have only internal IP addresses can access IP addresses for Google Cloud and services." Private Google Access is the specified subnet feature. Cloud NAT can provide internet egress but is not this subnet feature, peering links VPC networks, and a route alone does not enable internal-only VM access to Google APIs.

   Reference: Configure internet access and firewall rules — Access Google Cloud APIs

3. 3A telemetry store needs very high read/write throughput over petabytes of single-keyed records, with low-latency access by row key. Relational joins and multi-table SQL transactions are not required. Which listed service is designed for this access pattern?

   1. A. BigQuery
   2. B. Cloud Storage
   3. C. Cloud SQL
   4. D. Bigtable

   Answer: D

   Explanation: The guide states, "Bigtable is ideal for storing large amounts of single-keyed data with low latency." Bigtable's single-key low-latency model fits the telemetry access pattern. The warehouse, object and conventional relational stores do not provide that specified keyed wide-column architecture.

   Reference: Bigtable overview — Bigtable overview — Introduction

4. 4A set of BigQuery dashboards repeatedly queries frequently accessed data. The engineer wants an in-memory analysis service that can accelerate supported SQL queries without rewriting every dashboard query. Which feature fits?

   1. A. A materialized view
   2. B. The query result cache
   3. C. BI Engine
   4. D. A search index

   Answer: C

   Explanation: The guide states, "BigQuery BI Engine is a fast, in-memory analysis service that accelerates many SQL queries in BigQuery" BI Engine is the in-memory SQL analysis service. Materialized results, identical-query result reuse and search indexing are different acceleration mechanisms.

   Reference: Introduction to BI Engine — Introduction to BI Engine

5. 5Before running an on-demand BigQuery query on a table without row-level security, an engineer wants to validate the query and estimate bytes scanned without executing it. Which job configuration provides this?

   1. A. Set priority to BATCH and submit the query normally.
   2. B. Run SELECT * with LIMIT 1 as the cost estimate.
   3. C. Set dryRun to true in the query job configuration.
   4. D. Run the query once and rely on the next execution's result cache.

   Answer: C

   Explanation: The guide instructs, "submit a query job with dryRun set to true". Its dry-run example validates query structure and estimates scanned bytes. Batch priority, LIMIT, and a first executed run do not provide the requested non-executing estimate.

   Reference: Estimate and control costs — Perform a dry run — API

6. 6An analyst can read a BigQuery dataset but cannot query a column protected by an enforced policy tag. The analyst should receive read access to that sensitive column without administering its taxonomy. What additional access is needed?

   1. A. Data Catalog Policy Tag Admin on the taxonomy
   2. B. BigQuery Job User on the billing project
   3. C. Data Catalog Fine-Grained Reader on the policy tag
   4. D. Cloud KMS Viewer on the dataset encryption key

   Answer: C

   Explanation: The guide states, "The Data Catalog Fine-Grained Reader role is required for users who need access to data in secured columns." Dataset access does not replace this check. Taxonomy administration, job creation, and key metadata access do not grant the required fine-grained read access.

   Reference: Introduction to BigQuery column-level access control — Roles used with column-level access control

7. 7Dataflow workers have no external IP addresses and must call a third-party API on the public internet. Assuming routes and firewall rules permit it, which network service provides outbound internet access?

   1. A. Private Google Access alone
   2. B. VPC Network Peering alone
   3. C. Cloud NAT
   4. D. A default internet-gateway route alone

   Answer: C

   Explanation: The guide states, "This option is for running jobs that access APIs and services outside of Google Cloud that require internet access." Cloud NAT translates outbound traffic for internal-only workers. Google API access, peering, or an internet-gateway route alone does not supply the required third-party internet egress translation.

   Reference: Configure internet access and firewall rules — Internet access for Dataflow

8. 8An operational platform needs relational semantics, SQL, horizontal scaling, and strongly consistent transactions across regions. Which service combines these capabilities?

   1. A. BigQuery
   2. B. Spanner
   3. C. Cloud Storage
   4. D. Memorystore for Redis

   Answer: B

   Explanation: The Spanner page states, "A distributed clock called TrueTime guarantees transactions are strongly consistent even across regions." It also describes horizontal scaling and relational SQL semantics. A warehouse, object store, or Redis cache does not supply this operational relational transaction architecture.

   Reference: Spanner — How It Works

9. 9A BI Engine reservation uses preferred tables. A dashboard query joins two tables, but only one is in the preferred list. Assuming the SQL is otherwise supported, what list change is required for acceleration?

   1. A. Include both joined tables in the preferred list.
   2. B. Include only the larger joined table in the preferred list.
   3. C. Include only the first joined table in the preferred list.
   4. D. Keep a nonempty preferred list containing neither joined table.

   Answer: A

   Explanation: The guide states, "Queries that access multiple tables are only accelerated if all tables are preferred tables." All tables accessed by the query must be preferred. Each incomplete nonempty configuration fails this requirement; the question does not use the special empty-list case.

   Reference: Introduction to BI Engine — Preferred tables limitations

10. 10An on-demand BigQuery query must fail before execution if its estimated billed bytes exceed a specified per-query cap. Which control implements this requirement?

    1. A. Maximum bytes billed
    2. B. Project-level daily query usage quota
    3. C. A dry run without a maximum-bytes setting
    4. D. Batch query priority

    Answer: A

    Explanation: The guide states, "If the number of estimated bytes is beyond the limit, then the query fails without incurring a charge." Maximum bytes billed imposes the per-query estimated-byte cap. A daily aggregate quota has another scope, a dry run only estimates, and priority controls scheduling rather than this cap.

    Reference: Estimate and control costs — Restrict the number of bytes billed per query